00:00:00
I read a stat last week that there's
00:00:01
like a quarter of the amount of people
00:00:03
there's rats in New York City. Like
00:00:05
there's a quarter. Yeah. Wait, what?
00:00:07
>> Wait, hold on. I don't understand.
00:00:08
>> You're saying there's four times as many
00:00:09
rats in New York City.
00:00:11
>> One quarter as many quarter of a rat for
00:00:14
every four people.
00:00:15
>> There's one rat for every
00:00:17
>> Yeah, one rat. Isn't that That's like
00:00:18
>> That's a lot of rats. One rat for per
00:00:21
four person people.
00:00:22
>> So there's a 25% chance you have a rat
00:00:25
following you.
00:00:25
>> There's a 25% chance you are a rat.
00:00:34
Yo, what is up people of the internet?
00:00:35
Welcome back to another episode of the
00:00:37
Waveform Podcast. We're your hosts. I'm
00:00:39
Marquez.
00:00:40
>> I'm Andrew.
00:00:40
>> And I'm David.
00:00:42
>> And this week there's music playing for
00:00:43
some reason. I don't know why.
00:00:44
>> For some reason,
00:00:45
>> but it feels
00:00:46
>> We're all back. We're all back. We're
00:00:49
here.
00:00:50
>> Normal crew.
00:00:51
>> Spoiler. We didn't even say our names
00:00:53
yet.
00:00:54
>> We did. We did.
00:00:54
>> Audio didn't we?
00:00:56
>> Yeah, we did. We actually just You said
00:01:00
you
00:01:00
>> said I'm Andrew.
00:01:01
>> We went around and said our names and we
00:01:02
were like so we haven't said our names.
00:01:04
>> So we're all back. Uh we're all also
00:01:06
rusty, but hey,
00:01:08
>> it's uh it's another tech week. It's
00:01:09
Torch Tay Mech. It's Mech. Google IO
00:01:13
happened. Uh we got a whole bunch of
00:01:15
Gemini stuff. Sony tweeted they have
00:01:17
these new headphones, but they also
00:01:18
tweeted this April Fool's Day joke like
00:01:20
two months late. It was about AI photo
00:01:22
correction. We'll talk about that. Uh, a
00:01:24
new camera might be a Fuji X100V
00:01:27
competitor, maybe. And, uh, a story
00:01:29
about the coolest things we've ever done
00:01:30
as a company because it's it's new and
00:01:33
we've been keeping it a secret, but it's
00:01:35
time you guys finally know about this
00:01:36
thing that we've been doing. And I'm
00:01:38
very excited to share it with you.
00:01:39
>> At the end, can't skip ahead. Full
00:01:41
retention.
00:01:41
>> Yeah.
00:01:42
>> Wait, I I don't know what this is, so
00:01:43
I'm going to be excited.
00:01:44
>> I'm awesome. I was like, yeah, totally.
00:01:46
>> It's incredible.
00:01:47
>> Uh, totally. Yeah.
00:01:48
>> But first, did they even test this?
00:01:50
>> Yes. I think David has some that are for
00:01:53
next week that are going to be really
00:01:54
fun.
00:01:55
>> I'm a teaser.
00:01:56
>> I'm gonna give you guys a choice.
00:01:58
>> I was writing some things down on my my
00:02:00
fraternity. So, I have a few, but I have
00:02:03
either a did they even test this or one
00:02:05
that I was impressed by. So, I wrote it
00:02:06
as a hell yeah, they actually did test
00:02:08
this. So, people have been saying maybe
00:02:10
some of these things are a little too
00:02:11
downer. I I'll open the floor that if
00:02:13
you're ever really really impressed by
00:02:15
something, we'll put that in because
00:02:17
they did test it. I do like when things
00:02:19
surprisingly work and you didn't expect
00:02:21
them to.
00:02:22
>> Yes, this is one of those. It sounds
00:02:24
like we want to go with the hell yeah,
00:02:25
they tested.
00:02:26
>> I think that would be fun.
00:02:27
>> Cool. Sure.
00:02:28
>> This is something
00:02:29
>> that's positive. Worked really well.
00:02:31
It's something that maybe surprised me
00:02:33
because this feature hasn't worked for a
00:02:35
long time. Um, but I had a really good
00:02:38
magic Q experience
00:02:40
>> in Google Messages.
00:02:41
>> Remind what remind people what Magic Q
00:02:43
is. Magic Q is a feature in Google
00:02:46
Messages where depending on what the
00:02:48
person is sending you in a text message,
00:02:50
you'll get instead of like an auto reply
00:02:52
of yes, no, whatever, like a suggestion
00:02:54
of maybe they're asking about an event.
00:02:57
So here's your calendar and you can post
00:02:59
in the calendar event for it. It's
00:03:00
supposed to help you answer things that
00:03:02
Google should know within your Google
00:03:04
land easier. Yes.
00:03:06
>> So, I had somebody, a mutual friend of
00:03:09
Marquez and I say,
00:03:10
>> "Hey, do you have any photos of that
00:03:13
camera crane arm you used to have on
00:03:14
Marquez's Tesla?" And I was like, "Wow,
00:03:18
that was a long time ago. We don't have
00:03:19
that anymore. I'm not even sure. I'm
00:03:21
sure they're in here somewhere." And I
00:03:23
got a little button that said Google
00:03:25
Photos. I was like, "Oh, cool. I can one
00:03:27
click. I'm still going to have to search
00:03:28
for it."
00:03:28
>> Google Photos. launched Google Photos
00:03:31
right there. Suggestions based on the
00:03:33
text message and there's two pictures of
00:03:34
Marquez's car with the camera screen.
00:03:36
>> Wow.
00:03:37
>> And I just clicked both of them and
00:03:38
said, "Thank goodness for Magic Q
00:03:40
because these are actually them." And
00:03:42
Mtos had no idea what I was talking
00:03:44
about, but who cares? He got the
00:03:45
pictures he needed.
00:03:46
>> That is what it's supposed to be amazing
00:03:48
for.
00:03:49
>> Supposed to do. It is now 1 14 3000
00:03:54
>> that yeah it's I found that I can I can
00:03:56
search brand names in Google photos and
00:03:58
it can it can tell me like Tesla versus
00:04:01
regular car or you can search Mercedes
00:04:03
and it will find Mercedes.
00:04:04
>> Here's something that's funny. It said
00:04:07
on a Tesla like in the suggestion said
00:04:09
on a Tesla but
00:04:11
>> it was one of like eight photos that it
00:04:12
tested. The one before it was the um
00:04:16
trailer hitch camera mount with a giant
00:04:18
Ford logo on it. It always clearly is
00:04:21
not I don't know how, but it always has
00:04:22
false positives for some reason. But
00:04:24
yes, it did. It did find the Teslas,
00:04:26
which is cool.
00:04:27
>> Cool.
00:04:27
>> I think it found the crane arm. I think
00:04:29
that was the
00:04:30
>> the discerning factor that was a little
00:04:32
easier to pick up on and unique.
00:04:34
>> It's been getting the faces wrong for me
00:04:36
recently enough. Yeah. I'll be like this
00:04:38
person in Canada and it'll be like 80%
00:04:40
of them is that person, but then 20% of
00:04:43
them will be a different person.
00:04:44
>> False positives. Yeah. It just seems to
00:04:46
cast the net wider than it's supposed
00:04:47
to, which is probably a good miss.
00:04:49
>> Yeah.
00:04:50
>> And then Yeah. It'll just have some
00:04:51
random stuff that's not what you asked
00:04:52
for, but it will always have somewhere
00:04:54
in there one of the things you asked
00:04:55
for,
00:04:55
>> right? Just Okay. I tweeted that I'm
00:04:59
going to have a generational crash out.
00:05:01
So, I think that's about to happen.
00:05:03
>> And what will you be speaking out about
00:05:05
today, David? If you're unaware, this
00:05:08
week two companies decided to do the
00:05:11
worst thing possible to piss me off.
00:05:15
That's right.
00:05:17
Uh Google and Sony decided to implement
00:05:20
new AI image editing features.
00:05:23
>> That just proves that engineers have no
00:05:26
idea what the a photo is supposed to
00:05:27
look like.
00:05:28
>> I think they tagged you specifically
00:05:30
when they posted.
00:05:31
>> I felt personally very attacked by this.
00:05:34
>> I want to
00:05:35
>> Yeah. Yeah, before you get into it, the
00:05:36
blame, I'm not sure where to put it on
00:05:38
just the grouping of engineers cuz we've
00:05:40
had photo engineers come to us and say
00:05:43
like, "We watched your video on this and
00:05:45
we've been trying to tell people on our
00:05:47
team."
00:05:48
>> I'm blaming social media.
00:05:49
>> So, there's some there is someone to
00:05:52
blame and I agree and it doesn't take
00:05:53
anything away from this. Some of the
00:05:54
engineers know what they're talking
00:05:56
about.
00:05:56
>> Yeah. Okay. Well, to be fair, okay, I'm
00:05:58
just going to talk about exactly what
00:05:59
happened. Do you do you want to talk
00:06:00
about Sony first or Google first?
00:06:02
>> Let's do the Sony one.
00:06:03
>> The Sony one. Okay. So, Sony tweeted
00:06:06
this photo from a new phone that of
00:06:08
course they just basically didn't
00:06:09
announce like usual. Um, called the
00:06:12
Xperia, what is it? Xperia 1 Mark 8.
00:06:16
>> That's probably way too expensive and
00:06:18
they'll announce it now and it'll
00:06:19
release in September and then nobody
00:06:20
will buy it and it'll be available only
00:06:22
at B&H.
00:06:23
>> It's available for pre-order. Okay.
00:06:25
>> It's available for pre-order.
00:06:26
>> You can spend £1,400 on it and it ships
00:06:29
in a month in June.
00:06:30
>> In June. That's actually way faster than
00:06:32
they've ever done. They're getting
00:06:33
better. It's pretty bad, but it's better
00:06:35
than they've been in a long time.
00:06:36
>> Yes.
00:06:36
>> So, there there's what's happening.
00:06:38
>> So, they they tweet this tweet that
00:06:40
says, "The new AI camera assistant um
00:06:43
asterisk for some reason."
00:06:44
>> Uh with Experia intelligence brings
00:06:47
stories to life using subject, scene,
00:06:49
and weather. It suggests expressive
00:06:50
options with adjustments of color,
00:06:52
exposure, bokeh, and lens, whatever that
00:06:55
means, for breathtaking photos. What
00:06:57
does it mean? Adjustments for lens. What
00:06:59
does that even mean? Okay, there are
00:07:02
four photos here. One of them is kind of
00:07:05
acceptable, I suppose,
00:07:07
>> brother.
00:07:07
>> It's like a flower and they made it
00:07:09
warmer and it's like fine. The other two
00:07:12
are it's it's hard to explain how bad
00:07:15
these are. Um, one of them is almost
00:07:18
definitely not directly from the camera
00:07:20
because the shadows were already jacked
00:07:21
and it was already underexposed. And I'm
00:07:24
pretty sure that this was pre-edited,
00:07:26
but the AI edited version is like
00:07:28
sharpening to 12, uh, whites to 12,
00:07:31
saturation to -3. They boosted the red
00:07:35
for no reason.
00:07:35
>> To explain the photo, it's a woman
00:07:37
standing in a field full of like tall
00:07:39
hay and a blue background. And like you
00:07:41
said, it's a little underexposed.
00:07:42
>> And the horizon is at 30°.
00:07:44
>> Yeah.
00:07:45
>> How is she standing straight, but the
00:07:46
horizon?
00:07:47
>> That might be the more conf.
00:07:51
>> Is this a fake picture? I don't know.
00:07:53
Strangest social media post I've ever
00:07:54
seen.
00:07:54
>> Okay, but the the mo the worst one is a
00:07:57
sandwich.
00:07:58
>> The sandwich
00:07:59
>> which already like looked like a phone
00:08:01
photo. Like it has that compression. It
00:08:03
has that oversharpening. Whatever. At
00:08:04
least it had contrast. The AI camera
00:08:07
assistant one. One of the worst edits
00:08:10
I've seen in my entire life. Uh, I have
00:08:12
an Instagram post from before I knew
00:08:15
what cameras were where I applied like
00:08:17
three Instagram filters to make it look
00:08:19
kind of similar to this where it's just
00:08:21
like really washed out with like no
00:08:23
color whatsoever. Like I used to I used
00:08:25
to think that that's what made a good
00:08:27
picture too.
00:08:28
>> We all did some of those too content.
00:08:32
>> My bio since the the beginning of my
00:08:34
Instagram account was I'm I promise not
00:08:37
to overdo the filters. I mean, but did
00:08:38
you look at this?
00:08:39
>> # no filter was my my end of the
00:08:41
spectrum. Oh, yeah. That was
00:08:42
>> bad, right?
00:08:43
>> You're Yeah,
00:08:44
>> it's so bad. Guilty.
00:08:45
>> This This sandwich is like It's a
00:08:47
croissant with I don't know.
00:08:49
>> Turkey, cheese, tomato, lettuce, like a
00:08:52
sandwich.
00:08:53
>> Lots of good colors that could be Yeah,
00:08:54
it's a really close-up of a sandwich.
00:08:56
Yeah,
00:08:57
>> but it is so brightened
00:09:00
out in the new one that the white text
00:09:02
that says AI camera assistant is almost
00:09:04
hard to read in the photo because the
00:09:07
photo behind it is so bright and washed
00:09:10
out that it gets lost.
00:09:12
>> Yeah. Um, okay. So, if you look at
00:09:14
smartphone camera upgrades since like
00:09:15
the iPhone 7, specifically from US
00:09:18
smartphone manufacturers, you're
00:09:20
probably going to notice that there
00:09:22
hasn't been a lot of change over the
00:09:23
years. You know, the iPhone like 11, 12,
00:09:26
13, 14, 15, 16, 17. Like, we did these
00:09:30
camera tests, right? Smartphone cameras
00:09:32
over time, especially from US
00:09:35
manufacturers, really what they've
00:09:36
gotten better at is they've gotten
00:09:37
better at low light because the sensors
00:09:39
get bigger, they get better at the
00:09:41
stabilization, the low light eventually
00:09:44
gets better. However,
00:09:45
>> Mhm.
00:09:46
>> the primary thing that these companies
00:09:47
have been doing is making signs easier
00:09:50
to read because they look at user data
00:09:52
and they say, "What do people usually
00:09:53
use their smartphone for?" And it's not
00:09:56
to take artsy pictures, right? It's to
00:09:58
take pictures of signs to send
00:09:59
information to each other. A
00:10:00
>> lot of utility photos.
00:10:01
>> A lot of utility. That's
00:10:02
>> especially Zoom.
00:10:03
>> Yeah. Of It's probably like 80% of what
00:10:05
people take photos of, right?
00:10:06
>> Depends on the age.
00:10:08
>> Depends on the age.
00:10:08
>> I think a lot of old people have a lot
00:10:10
of utility photos.
00:10:11
>> That is very
00:10:12
>> You're not taking pictures of kids
00:10:13
anymore. It's just like menus at
00:10:15
restaurants, signs to read them better.
00:10:17
>> I mean, I take photos of those things,
00:10:19
send them to friends, whatever.
00:10:21
>> Um, but people are buying dig cams right
00:10:23
now because all these smartphone
00:10:25
manufacturers have just been jacking up
00:10:27
the shadows, crushing the highlights,
00:10:29
adding sharpening, you know, like, oh,
00:10:31
people are buying dig cams because
00:10:33
they're old and they're like
00:10:34
fashionable. It's like, no, the dig cams
00:10:36
that people are buying now are still
00:10:38
good cameras. It's just that they don't
00:10:40
apply a ton of computational photography
00:10:42
to these camera to these photos, right?
00:10:44
>> Mhm.
00:10:45
>> So, I have a diagram for you guys that
00:10:46
I'm going to pull up. It's not a
00:10:47
diagram, but it's something I made this
00:10:48
morning.
00:10:49
>> It's just a bunch of red yarn
00:10:52
strewn across.
00:10:53
>> So, this is a normal picture. This is a
00:10:55
picture I took in Colorado a couple
00:10:56
weeks ago, right?
00:10:57
>> Mhm.
00:10:57
>> It is a mountain landscape. There's some
00:11:00
trees in there. There's some bright
00:11:01
clouds. There's snow, which is quite
00:11:02
bright. Then there's some, you know,
00:11:04
there's a lake, stuff like that.
00:11:05
>> Blue sky.
00:11:06
>> Blue sky. So, there's kind of a lot
00:11:07
going on. For people that don't know,
00:11:09
there's a thing called a histogram,
00:11:11
which has a thing in it called a tone
00:11:12
curve. And effectively, it's a it's a
00:11:15
chart that moves from left to right and
00:11:17
up and down. And on the left side,
00:11:19
you've got your shadows, in the middle,
00:11:20
you got your midtones, and on the right
00:11:22
you've got your highlights. And to have
00:11:24
a wellbalanced histogram, it's kind of
00:11:26
important to have information in the sha
00:11:29
like to have a lot represented in the
00:11:31
shadows, mid tones, and highlights. What
00:11:33
these smartphone camera companies are
00:11:35
doing
00:11:36
is they are jacking the out of the
00:11:39
shadows. They are crushing the
00:11:41
highlights. They are adding clarity. Uh
00:11:44
I would say a negative dehaze
00:11:46
specifically on that sandwich photo.
00:11:48
>> Yeah.
00:11:48
>> And it looks like this. And this is kind
00:11:50
of what that AI camera assistant photo
00:11:51
looks like. And if you look at the
00:11:52
histogram, they pull in all of the
00:11:55
highlight information and the shadow
00:11:57
information and it all goes away and it
00:11:59
all gets crushed into the center.
00:12:00
>> It's the middle. So, it's one big
00:12:01
>> It's all a midtone. Everything is a
00:12:03
midtone.
00:12:04
>> Mhm.
00:12:04
>> Would you say these smartphone companies
00:12:06
should atone for their tone curve?
00:12:08
>> Yeah. Yes. They need to atone for their
00:12:10
sins.
00:12:10
>> Thank you.
00:12:11
>> Yeah. I'm back, baby.
00:12:12
>> I just want to say, and we've made this
00:12:14
analogy before, the difference between a
00:12:16
circle and a sphere is shadow is shadow
00:12:21
is is like curvature, which you can only
00:12:23
add through like a transition from
00:12:26
highlight to shadow. Right. Carvajio,
00:12:29
Da Vinci, they died for this. Okay, they
00:12:32
invented this. Paintings in Italy in the
00:12:35
Middle Ages were all flat as and then
00:12:37
they were like, "What if we added
00:12:39
transitions between color and and tone?"
00:12:41
And then things felt realistic and
00:12:44
three-dimensional and people are like,
00:12:45
"Wow, we can have realistic stuff." And
00:12:47
all these smartphone companies are just
00:12:49
like, "Nah, bro. We're going back to
00:12:50
before the Middle Ages."
00:12:52
>> Yeah.
00:12:52
>> And I I don't know. I just I there's
00:12:55
something like the the Chinese the
00:12:57
Chinese manufacturers are actually
00:12:59
making good cameras, but for some reason
00:13:02
the US manufacturers, they're just like,
00:13:04
"No, no, we're going to keep doing the
00:13:06
same computational photography that we
00:13:07
did when our when our sensors were like
00:13:09
a micron big."
00:13:10
>> I don't think it's just the US
00:13:11
manufacturers. I just I so I I was going
00:13:13
to say I talked about this in the big
00:13:14
Android update video because that that
00:13:16
before and after that they did,
00:13:18
>> right, which we'll get to a second. And
00:13:19
I talked about processing and I I did
00:13:21
this thing where I had a every single
00:13:23
version of a camera of every single
00:13:25
iPhone, every single Samsung Galaxy S,
00:13:27
every single uh Google phone.
00:13:29
>> Yeah.
00:13:29
>> And I thought Samsung's phones were
00:13:31
pretty strong. Like the HDR happening in
00:13:33
the S26s versus like the S22s and S23s,
00:13:36
it looks worse because they're trying to
00:13:38
do this thing where they save bad
00:13:40
photos. Yeah.
00:13:41
>> Which feels like a win, but it also
00:13:43
overprocesses good photos, which is
00:13:46
definitely not a win.
00:13:47
>> Yeah. So yeah, that's S Sony, Samsung,
00:13:50
Google, Apple, a lot of them are guilty
00:13:52
of this in a lot of ways.
00:13:53
>> I mean, iPhone 7 and before the photos
00:13:55
look great because they actually had
00:13:57
contrast
00:13:57
>> in some ways. And they've gotten better,
00:13:59
obviously, too. You have more
00:14:00
resolution, you have more depth of
00:14:01
field, you have also things that have
00:14:02
gotten, you know, more impressive at
00:14:04
capturing and getting good photos out of
00:14:05
like bad situations. But yeah,
00:14:07
>> they just need to change how they
00:14:08
process the information now that they
00:14:10
have way bigger sensors. They haven't
00:14:11
done that. Like they're using the same
00:14:12
algorithms that they were using on the
00:14:14
Pixel 1. You know,
00:14:15
>> you know a subreddit that's aged really
00:14:17
well.
00:14:18
>> What?
00:14:18
>> R/ HDR?
00:14:21
>> I subscribed to that subreddit a long
00:14:22
time ago because it was hilarious
00:14:24
because HDR was like this funny novel
00:14:26
thing that used to like used to be able
00:14:28
to overdo and you'd get uh pictures on
00:14:31
like Facebook like this.
00:14:32
>> Yeah.
00:14:33
>> Oh my god.
00:14:34
>> Hilariously overdone HDR.
00:14:36
>> What is it?
00:14:36
>> But but you can you get this effect.
00:14:39
This is just one of many extremely
00:14:41
overprocessed photos. when you drag
00:14:43
highlights all the way down, shadows all
00:14:45
the way up, clarity all the way up,
00:14:46
sharpness all the way up, that's what
00:14:48
you get. And that kind of feels like
00:14:51
>> there was an era in fine art
00:14:52
photography, like right after Photoshop
00:14:55
got invented, where people invented
00:14:56
bracketing, which is where you take on a
00:14:59
tripod, you take like an underexposed
00:15:00
photo, a perfectly exposed photo, and an
00:15:02
overexposed photo, and you use the
00:15:04
information from all of those three
00:15:06
photos or up to like nine to basically
00:15:08
make like a hyper HDR image. And there
00:15:11
was a trend like in the in fine art
00:15:13
photography specifically because it was
00:15:15
hard to do, right? What people consider
00:15:17
a good quality is basically things that
00:15:19
are difficult. So at the time you had to
00:15:21
have the knowledge to do this. And so if
00:15:22
you go into any like coffee shop in like
00:15:24
a fancy town, there's a giant photo on
00:15:27
the wall of like a mountain landscape or
00:15:28
a river that just looks horrible.
00:15:31
>> Yeah. Yeah.
00:15:32
>> I think
00:15:32
>> and it's always printed on metal. So
00:15:34
it's like it's like glossy and weird. I
00:15:37
hate that.
00:15:38
>> Oh my god. I'm so glad you said that.
00:15:40
>> Any like gallery in a town outside a
00:15:43
national park.
00:15:44
>> Yes. Like Yeah. Yeah. It's just like a
00:15:46
bunch of animal photos that were taken
00:15:48
with great cameras and like
00:15:49
>> it's almost like I need to make this as
00:15:53
>> cool and like
00:15:54
>> impressive as possible cuz I took this
00:15:56
amazing picture of a moose, but it's
00:15:57
just like
00:15:57
>> and they sell them for like $5,000.
00:15:59
>> It'll be like a sunset where the sky is
00:16:01
like so orange. It's like you melted
00:16:04
10,000 basketballs
00:16:06
and like a wolf that's like silhouetted
00:16:09
but because it's bracketed there's still
00:16:11
tons of Sorry. This is like my pet
00:16:12
peeve.
00:16:14
>> Was that near Estus Park? There's
00:16:16
probably 40 galleries in Estus Park that
00:16:18
have like these exact pictures I'm
00:16:19
talking I specifically remember one next
00:16:21
to a coffee shop. Yeah, I think there is
00:16:23
something to what you said where like
00:16:24
people think that it's good if it looks
00:16:26
difficult and if there's like back in
00:16:29
the day when photos were 8 megapixels if
00:16:30
you get a really sharp photo that would
00:16:32
look like a really good camera. So if
00:16:34
you could
00:16:35
>> sharpen dehaze and like create a really
00:16:38
you know structured sharp photo it
00:16:40
looked quote better even though it
00:16:41
didn't really look that good
00:16:42
>> just cuz you didn't know how to do it.
00:16:44
>> Yeah. So that sort of stuff started
00:16:45
leaking into like this generic
00:16:48
photography. It Yeah, that sort of
00:16:50
bracketing is is the same thing. And
00:16:52
it's like, okay, if I want to take a
00:16:53
picture with like the blue sky in
00:16:55
instead of blown out, it would be really
00:16:57
difficult unless I had a photo a camera
00:16:59
with great dynamic range or I could
00:17:00
learn bracketing and get like two or
00:17:02
three photos and combine them.
00:17:03
>> Yeah.
00:17:04
>> So now every time I take a photo, I'm
00:17:05
like, I should do bracketing just to
00:17:07
make sure it looks better. It's not not
00:17:09
necessarily better, but
00:17:10
>> then that starts creeping into regular
00:17:12
photos, too.
00:17:12
>> Totally. And this has even happened in
00:17:14
film photography. Like there's a film
00:17:16
photography is kind of trendy right now,
00:17:17
you know, whatever. There's this like
00:17:19
saying that people will be like, "Oh,
00:17:21
just always overexpose by by a stop so
00:17:24
that you always have information in the
00:17:25
shadows." And it's like,
00:17:27
>> "No, because then you're just gonna get
00:17:29
an image that has too much information
00:17:31
in the shadows." And like,
00:17:33
>> yeah, I guess you can scale it back
00:17:34
because the highlights are more
00:17:35
protected, but it's better to actually
00:17:37
expose for the photo you want, not just
00:17:39
show all the information all of the
00:17:40
time.
00:17:41
>> Yeah.
00:17:41
>> You know, so to get back to the Sony
00:17:43
thing,
00:17:44
>> yeah, they Sorry. Well, two things on
00:17:46
the Sony thing. one, that tweet with all
00:17:48
the horrible edits from the AI thing
00:17:51
blew up. It has 13 million views on
00:17:54
Twitter.
00:17:55
>> Their marketing department is like, I'm
00:17:57
so confused because the best time to
00:17:58
delete this was yesterday, but also this
00:18:01
is the most traction we've ever gotten.
00:18:03
>> It's a lot of views. So, on that
00:18:05
subject, the second thing is they posted
00:18:07
a follow-up tweet.
00:18:08
>> Yeah.
00:18:08
>> Uh that said, did following the post
00:18:11
about the AI camera assistant, we'd like
00:18:12
to explain the feature in more detail.
00:18:14
It doesn't edit photos after shooting.
00:18:17
It suggests four settings in different
00:18:19
creative directions based on the scene
00:18:21
and subject. You can choose any option
00:18:24
and use your own or use your own
00:18:26
settings. And so they show a couple more
00:18:28
examples of one original photo and then
00:18:31
four suggested AI examples.
00:18:33
>> Yeah.
00:18:33
>> Which almost makes it more baffling that
00:18:36
they decided to show such a horrible
00:18:38
example of a creative direction on all
00:18:41
of those photos. They look dramatically
00:18:42
worse. Yeah,
00:18:43
>> they thought that was creative
00:18:45
direction, I guess, is a choice.
00:18:47
>> To be fair, two of these suggestions out
00:18:49
of the four are horrible, too.
00:18:51
>> But they're nowhere near as bad as how
00:18:53
bad the other ones were. So like even in
00:18:55
this has four suggestions and none of
00:18:57
them are as bad, whereas the other one
00:18:59
>> that went through approval that those
00:19:02
>> dude so many things about this like even
00:19:04
just the the base photo looks like it's
00:19:06
like one megapixel. Like it's so
00:19:08
oversharpened and soft and
00:19:11
>> Yeah. Do you think they'd sue us if we
00:19:13
sold a t-shirt with that sandwich photo?
00:19:15
Cuz I would
00:19:16
>> I also love that you can barely read the
00:19:18
text that says, "I camera assistant
00:19:20
original on it." Yeah.
00:19:21
>> So anyway, they the social media post
00:19:23
was baffling. Whoever decided to choose
00:19:26
those as the examples for the post ended
00:19:28
up winning because they got 13 million
00:19:30
post views on their social media post.
00:19:32
But also like now everyone's dunking on
00:19:34
this Sony phone again, which might be
00:19:35
the most attention any Sony Xperia 1 has
00:19:36
gotten in a decade.
00:19:38
>> Yeah, but that's true.
00:19:39
>> Maybe not for the right reason. I mean,
00:19:40
I probably wouldn't even have heard that
00:19:42
they had released this. I didn't even
00:19:43
know they released the the last one.
00:19:44
>> That's actually true. I didn't know that
00:19:46
this phone was coming out.
00:19:46
>> You know what? How I knew it was coming
00:19:48
out like 2 days before this. Someone on
00:19:50
our subreddit was like, "Why does big
00:19:52
tech reviewers never do Sony phones?
00:19:54
This one's coming out and it's clearly
00:19:56
going to be the best camera. It would
00:19:57
automatically win the smartphone award."
00:19:59
And then this came out and a bunch of
00:20:00
people in the comments were like, "Is
00:20:02
this the camera you were talking about?"
00:20:04
>> Maybe not, my friend. Hey, the
00:20:05
smartphone award. I don't know.
00:20:07
>> Okay. And another thing happened with
00:20:09
the camera editing this week. Um, so I
00:20:12
was I missed the Android show episode
00:20:13
and I watched the Android show while I
00:20:15
was in Europe and it was mostly good.
00:20:17
You know, I actually liked a lot of it.
00:20:19
I thought Google did a good job kind of
00:20:20
humanizing their um their upper
00:20:22
management people.
00:20:24
>> Oh, I hard disagree.
00:20:25
>> I thought it was pretty
00:20:26
>> cringy, which was very human. It was
00:20:28
cringey. Really cringey. It was It was
00:20:30
cringey.
00:20:31
>> Every time it made me face palm like
00:20:33
that's that was the vibe. the like the
00:20:35
like half cut of like am I supposed to
00:20:37
say something funny here? Like oh boy.
00:20:40
>> I didn't watch it but that sounds
00:20:42
brutal.
00:20:42
>> I think just the way
00:20:43
>> before they talk they're like
00:20:45
>> they'd have like an unscripted moment
00:20:47
which was
00:20:47
>> which was probably scripted. Yeah.
00:20:49
>> Yeah. It was tough.
00:20:50
>> I think if you have to describe it as
00:20:51
humanizing upper level management
00:20:55
red flag.
00:20:56
>> That's true.
00:20:57
>> It's hard for all these companies. They
00:20:58
have like business leaders and like
00:21:00
ruthless engineers like trying to also
00:21:02
be the person who presents the thing on
00:21:03
camera. I mean, IOIO is rough. Mine is
00:21:05
Woodward. He was great. But everyone,
00:21:08
almost every company does a terrible job
00:21:09
at it. I think Apple's the gold standard
00:21:11
that they're all trying to match because
00:21:13
they have these incredibly Yeah. robot
00:21:16
energy, but they at least have like way
00:21:18
to produce conver uh conversational
00:21:20
types of presentations, but they also
00:21:21
don't pretend like they're actually
00:21:24
having a real conversation. They're not
00:21:26
like, "Hey, Jason, would you mind if you
00:21:29
sent me that screenshot of this?" Sure,
00:21:31
Cheryl.
00:21:32
>> Like, they don't do that kind of thing.
00:21:33
Honestly, since co it was like when we
00:21:35
had real live technology in your house
00:21:38
live presentations, those were that was
00:21:41
like a an era and then after co we got
00:21:44
these like produced things. That's a
00:21:46
this new era that we're in and they're
00:21:48
trying to make this produced era of
00:21:50
presentations feel kind of real again by
00:21:54
having unscripted moments or like long
00:21:56
one takes or Carl Pay just like not
00:21:59
blinking for five minutes or whatever
00:22:00
they got to do to make it feel like a
00:22:02
person. Yeah, that
00:22:03
>> well and also because the the general
00:22:05
sentiment around technology is at an
00:22:06
all-time low right now.
00:22:08
>> Like for regular people, regular people
00:22:10
hate technology right now.
00:22:11
>> Yeah. And they're trying to put the
00:22:13
person there. So you feel like you're
00:22:14
talking to a person instead of getting a
00:22:16
tech demo at IO, which is a giant AI
00:22:18
show.
00:22:19
>> Just to go on everyone, the resentment
00:22:20
is like, have you seen all these uh
00:22:22
commencement speeches getting booed?
00:22:24
>> Yeah. Getting booed like at all the got
00:22:27
booed like crazy. Yeah.
00:22:29
>> There's been like multiples of them in
00:22:30
the last week. So
00:22:31
>> Yeah.
00:22:32
You're going to speak to a bunch of
00:22:34
students who are graduating like
00:22:35
metriculating into the next part of
00:22:37
their life and you're going to talk
00:22:38
about agentic force. It's like what's
00:22:40
wrong with you?
00:22:42
>> That's insane to choose to talk.
00:22:43
>> Yeah. It's a little it's a little
00:22:45
>> that. But so
00:22:48
I think I missed this photo you're about
00:22:50
to talk about.
00:22:50
>> Oh, really?
00:22:51
>> Yeah. Yeah. I haven't seen it, so I'm
00:22:52
excited.
00:22:52
>> You haven't seen it before and after?
00:22:54
>> I have not seen.
00:22:54
>> I tweeted it. I tweeted it. I'll show
00:22:56
you in.
00:22:56
>> Okay. So, most of the features they
00:22:59
released at the Android show were
00:23:00
actually pretty cool. like they had the
00:23:01
um they have the widgets that you can
00:23:03
just generate on in in an instant and
00:23:06
kind of create whatever you want.
00:23:08
>> But one of them was basically the same
00:23:10
thing as this Sony thing except it was
00:23:12
like now you can take any social media
00:23:15
like photo or video and instantly make
00:23:17
it look better with like this new Google
00:23:20
feature which I thought is I thought
00:23:22
we've had auto mode for like a really
00:23:24
really long time so I'm surprised that
00:23:25
they considered this like new and
00:23:26
better.
00:23:27
>> Mhm. But they took a photo that looked
00:23:29
pretty decently well exposed. Like maybe
00:23:31
it was slightly underexposed a little
00:23:33
bit and then they did the exact same
00:23:35
thing that Sony did where they just like
00:23:37
made it way brighter. Uh softened it and
00:23:41
that was kind of it. Like they dragged
00:23:43
the exposure slider up a little bit but
00:23:45
they also jack jacked up the shadows and
00:23:47
pulled down the highlights. And it's
00:23:48
like
00:23:48
>> it would be a prime r candidate.
00:23:52
>> Yeah. And she's like, "It's so awesome
00:23:54
because you can take a photo that looked
00:23:56
like this and it's like a normal looking
00:23:57
picture." Like I would pref I would that
00:23:59
would be a fine photo in my opinion to
00:24:01
this and like the resulting photo is
00:24:03
just so bad looking.
00:24:05
>> Yeah. Now the thing is we know that
00:24:08
people prefer brighter photos especially
00:24:10
in this before and after context where
00:24:12
if you just had that original photo
00:24:14
they'd go, "Oh yeah, it looks fine." But
00:24:16
for some reason, regular people look at
00:24:19
their before and then look at the after
00:24:21
and go, "Oh, the after is better. It's
00:24:22
brighter. I can see more things.
00:24:24
Therefore, it's a better photo." And we
00:24:26
haven't been able to escape this. So,
00:24:27
these companies just keep seeing that
00:24:29
response that people have and just keep
00:24:32
doing what people keep rewarding, which
00:24:34
is making the photo brighter, seeing
00:24:35
more in the shadows, seeing more in the
00:24:37
highlights, and that's the result. I
00:24:39
have a theory that that people who are
00:24:42
not like art like are not creative
00:24:44
people who are not artists, they don't
00:24:46
have taste, but they like things that
00:24:48
are bad
00:24:50
>> because their taste is associated
00:24:52
>> I was going a different way.
00:24:54
>> What do you think?
00:24:55
>> I thought you were going to say they
00:24:56
have no taste, so they just try to
00:24:57
execute the technically better thing
00:24:59
even though it's taste.
00:25:00
>> No, I just like David telling them their
00:25:02
taste sucks.
00:25:03
>> But you said they don't have like things
00:25:05
that are bad,
00:25:05
>> but they like things that are bad. Well,
00:25:08
you know how you know how over time when
00:25:10
you get better at something things that
00:25:11
you you used to like look at and aspire
00:25:13
for you're like eventually like h
00:25:15
actually that wasn't that good and like
00:25:16
your your taste kind of moves upwards.
00:25:18
>> Yes.
00:25:18
>> And you you move around you're like oh
00:25:20
now I'm interested in this like I'm
00:25:21
looking up at this and it changes. I
00:25:24
think when you know if you're not in
00:25:26
that specific hobby or that specific art
00:25:28
form or that niche, you don't and you
00:25:30
don't really know what makes something
00:25:32
look good, you're gonna gravitate
00:25:34
towards the things that you think look
00:25:36
good at that time, right?
00:25:38
>> Yeah.
00:25:38
>> Should we just cut this whole section?
00:25:39
No.
00:25:41
>> Not subjective. And
00:25:42
>> yeah, I think it's
00:25:43
>> when I picture someone being an apple
00:25:45
right now.
00:25:46
>> No, I think it's it's an fascinating
00:25:47
conversation. When I picture someone
00:25:48
with no taste, I picture them not having
00:25:50
any direction at all to what they tend
00:25:52
to pick. But I think what happens is
00:25:53
regular people just see brighter as
00:25:56
better. The same way that people hear
00:25:58
louder as higher quality. It's just sort
00:26:00
of default. It's a default that humans
00:26:02
tend to just
00:26:03
>> just do that. And so now it's they don't
00:26:06
have taste, but that's the one thing
00:26:07
that they understand is better.
00:26:08
>> Right.
00:26:09
>> Also, if like I think for a lot of
00:26:11
people, they're not even looking at the
00:26:12
quality of their photo and they just
00:26:14
don't want the auto camera settings to
00:26:16
fail ever. Like the idea that like every
00:26:18
single picture comes out usable to some
00:26:21
degree I think is like a
00:26:22
>> Yeah, for sure.
00:26:23
>> I think it goes all the way back. We,
00:26:26
you know, we big circled this into David
00:26:29
said most people are looking at these as
00:26:30
tools. These are tools. The brighter
00:26:32
photo shows more information. The louder
00:26:33
thing I might be able to hear more of.
00:26:35
And most people care about it as tools
00:26:37
because they don't
00:26:38
>> mind that it's like a beautiful photo
00:26:40
you might hang. They mind that it's a
00:26:43
photo that they can see their kid having
00:26:44
fun on the playground.
00:26:45
>> Yeah. And to play devil's advocate for
00:26:47
the for the terrible HDR photo that we
00:26:49
get from the before and after,
00:26:50
>> it is easier for most people to turn the
00:26:53
after into what they want than the
00:26:55
before.
00:26:56
>> Does that make sense? I can turn up
00:26:58
contrast, bring the shadows back down,
00:27:01
and like make it a regular looking photo
00:27:03
out of the insane looking thing more
00:27:06
easily than someone looking at the
00:27:08
original, turning it into something that
00:27:09
they think looks good and bright.
00:27:10
>> Depends on how compressed they make it
00:27:12
though.
00:27:12
>> Definitely. But like in general like
00:27:14
people don't know how to make edit good
00:27:17
photos.
00:27:17
>> That's fair.
00:27:18
>> So it's easier to just give them a a
00:27:20
flat bright thing and be like, "Oh,
00:27:21
okay." Yeah. Oh, David says I should
00:27:23
have more shadows. I can hear. Turn it
00:27:24
down. Now there's shadows.
00:27:26
>> I want to make one more analogy.
00:27:27
>> Yeah.
00:27:28
>> You know, you know live love core stuff.
00:27:31
>> Yes.
00:27:31
>> Like that you'd buy at like Marshalls or
00:27:34
HomeGoods.
00:27:34
>> Homegoods or any cheap Airbnb you've
00:27:37
ever been in?
00:27:37
>> Every time.
00:27:38
>> David, you came back with some heat this
00:27:39
week.
00:27:39
>> I'm sorry.
00:27:41
>> We're going after Airbnb. All I'm saying
00:27:42
is like actual like interior designers
00:27:45
would never use that.
00:27:47
>> Interesting.
00:27:47
>> I don't know. This is a this could be a
00:27:49
whole
00:27:50
>> I'm trying I'm thinking of it as like
00:27:51
Yeah. It's like lines on a on a graph.
00:27:53
Like no taste is just no correlation at
00:27:56
all. It's just a blob.
00:27:57
>> But you're saying because they're being
00:27:58
gravitated,
00:27:59
>> but people are gravitating towards
00:28:01
something. Yeah.
00:28:02
>> Yeah.
00:28:02
>> So it's it's it's more than just no
00:28:04
taste. It's like some weird uh universal
00:28:08
constant that we all enjoy. It's
00:28:10
>> I don't I mean Yeah, you enjoy it until
00:28:12
you realize how bad it is.
00:28:13
>> Yeah. Yeah. And then you're gravitated
00:28:15
towards something else after that.
00:28:16
Exactly.
00:28:16
>> It's the first step. Maybe it's like a
00:28:18
step.
00:28:19
>> Yeah, it is. I think it is kind of a
00:28:20
ladder. I don't know. Anyway,
00:28:22
>> the taste ladder.
00:28:23
>> The taste ladder.
00:28:24
>> I just I just don't understand why
00:28:26
Google is is positioning this as like a
00:28:28
brand new like technological advancement
00:28:30
thing. It's like it's literally just
00:28:32
they hit the auto button.
00:28:34
>> Yeah.
00:28:34
>> Which has existed for like 25 years.
00:28:37
Like I don't
00:28:37
>> Yeah, but they're all AI now. They all
00:28:39
have to think.
00:28:40
>> Sure. I I don't
00:28:41
>> It's the artificial intelligence auto
00:28:43
button.
00:28:44
>> It's so weird.
00:28:45
>> Yeah,
00:28:45
>> it's so weird. Anyway, I just go try out
00:28:48
like the Leica Xiaomi phone or something
00:28:51
and you will see what a what a good
00:28:53
smartphone camera is.
00:28:54
>> It just comes back to what we were
00:28:55
saying before. Most of these
00:28:56
>> if it needs a smoke.
00:28:58
>> Really,
00:29:00
>> people just use these as tools. Like,
00:29:02
>> yeah,
00:29:03
>> we care about getting a good picture.
00:29:05
Most people want to capture the memory.
00:29:07
Well, that's ironic because the number
00:29:09
one seller of smartphones has always
00:29:11
been the camera.
00:29:12
>> Yeah. But not because they want to post
00:29:14
it on Instagram because they want to
00:29:16
look at it in 12 years and be like,
00:29:17
"Look at what my nephew was doing."
00:29:19
>> It is funny though because these events
00:29:22
are pointing towards people like us who
00:29:24
are judging the cameras based on taking
00:29:26
a nice photo. So for the people
00:29:28
>> That's true.
00:29:28
>> that works for them aren't watching the
00:29:30
Android show. The people who watch the
00:29:32
Android show think this sucks.
00:29:33
>> That's true. There's a there's a good
00:29:34
amount of comments of people being like,
00:29:36
"Oo, tech reviewers are so out of touch.
00:29:38
They always want this, that, and the
00:29:39
other." But really, regular people just
00:29:40
want like to zoom in to 50x and capture
00:29:43
their kid on a stage in kindergarten
00:29:46
being a pumpkin for Halloween. Like,
00:29:48
that's as long as you can get that and
00:29:49
it's clear, it's like great. That's a
00:29:51
win. It's like a screwdriver, not a
00:29:53
brush, you know?
00:29:54
>> I think I think it's beyond like people
00:29:56
not having taste. I just think most
00:29:57
people don't think critically and don't
00:30:00
care,
00:30:01
>> you know, about how things like look or
00:30:02
feel. Like remember like
00:30:04
>> you know this is this is the country
00:30:06
where we have about the same number of
00:30:08
strip malls as schools you know like
00:30:10
we're not thinking too deep about our
00:30:12
surroundings.
00:30:13
>> Yeah. There's way more places that I
00:30:14
could shop at. Why are do we have this
00:30:16
many schools?
00:30:17
>> Yeah. Exactly. You know we're like why
00:30:19
should anyone need to like you know
00:30:21
>> my fourth homes.
00:30:23
>> I haven't been to a school in forever.
00:30:25
>> Don't anymore.
00:30:26
>> Only like 10% of the population uses
00:30:28
schools.
00:30:29
>> Do you think there's more babies than
00:30:31
not babies? Yeah. Why do we have those
00:30:33
things?
00:30:34
>> I read a stat last week that there's
00:30:36
like a quarter of the amount of people
00:30:38
there's rats in New York City. Like
00:30:39
there's a quarter. Yeah. Wait, what?
00:30:41
>> Wait, hold on. I don't understand.
00:30:43
>> You're saying there's four times as many
00:30:44
rats in New York City.
00:30:45
>> One quarter as many quarter of a rat.
00:30:48
>> There's one rat for every four people.
00:30:50
>> There's one rat for every
00:30:51
>> Yeah, one rat. Isn't that That's like
00:30:53
>> That's a lot of rats.
00:30:55
>> One rat for per four person people.
00:30:57
>> So, there's a 25% chance you have a rat
00:30:59
following you. There's a 25% chance you
00:31:01
are a rat.
00:31:03
>> All right.
00:31:04
>> Men in black.
00:31:05
>> My face just comes off and it's a rat.
00:31:07
>> Yeah. Anyway, I understand these are
00:31:09
tools and brighter people think is
00:31:12
better. And it's the same reason why,
00:31:14
you know, Apple can be like, "Oh, we are
00:31:16
using pixel binning now." And so it's
00:31:17
the same resol, it's like the same
00:31:19
quality even though you're they're
00:31:21
getting less light per pixel. And then
00:31:24
dumb dumb asses like me will be like,
00:31:26
"Well, technically um the photo site is
00:31:29
quite a bit smaller." And so the quality
00:31:31
of the photo site is and my uncle or my
00:31:34
I don't even have an uncle. My freaking
00:31:36
I don't know. My grandma, she's dead. My
00:31:39
mom my mom will be like, I just want to
00:31:42
zoom in at 50x.
00:31:44
>> Yeah.
00:31:44
>> And that's all she wants to do.
00:31:45
>> Yeah. That'll Yeah. Like when we when we
00:31:47
see Samsung's space zoom and the first
00:31:50
thing we do is go 50x photos look bad
00:31:53
and then regular people see the
00:31:55
commercial and they're like finally I
00:31:56
can zoom into 50x and get that picture
00:31:58
of that sign or that random stage
00:32:01
performance for my kid that I was trying
00:32:02
to get a picture of.
00:32:03
>> Exactly.
00:32:03
>> Or like I'm in the crowd at the baseball
00:32:05
game and I can zoom all the way in on
00:32:06
their face for some reason.
00:32:08
>> Yeah, that's fair. Like the default is
00:32:10
going to be the one that most people use
00:32:11
and then the specialized tools are in
00:32:13
the marketplace of ideas. That's fair. I
00:32:16
do think Apple has a real opportunity to
00:32:18
release like a pro camera app for
00:32:21
themselves because they're really
00:32:22
getting their lunch eaten and they have
00:32:24
the opportunity to Sherlock like never
00:32:26
before.
00:32:27
>> So, we'll see if that happens.
00:32:29
>> True. Anyway,
00:32:31
>> I said I was going to have a
00:32:33
generational crash out. So, you can't
00:32:35
>> Well, you told most people they have no
00:32:36
taste. So, I think check the box.
00:32:39
>> Okay. By most people, yeah, I do mean
00:32:41
most people. I I mean,
00:32:42
>> no one listening to this though.
00:32:43
Everyone here.
00:32:44
>> All of you guys are good though.
00:32:46
Yeah, I'm going to seem like such a in
00:32:48
this episode.
00:32:49
>> Now, we have to establish that we are
00:32:51
better than the average person like you
00:32:53
>> by getting zero trivia questions
00:32:55
correct.
00:32:56
>> If I threw a Frisbee, you'd probably
00:32:57
think, "Oh my god, his form is
00:33:00
>> No, no, no, no. I would never judge a
00:33:01
That's a perfect analogy actually
00:33:04
because I know that you're not a Frisbee
00:33:06
player." Yeah. It's like a photographer
00:33:08
judging a random person's photo. I'm not
00:33:10
going to jud you're a random person. I
00:33:11
wouldn't judge but another photographer,
00:33:14
>> right? Like if if you were on a frisbee
00:33:16
team,
00:33:17
>> I would judge your form.
00:33:18
>> That's the analogy.
00:33:19
>> It can bite you in the ass, man. There
00:33:20
are so many warm-ups where I was like,
00:33:22
>> "Is that how that guy throws a forehand?
00:33:24
I'm going to tool him and get whooped by
00:33:26
them on the field later." I thought that
00:33:28
about so many people that are now really
00:33:30
good friends and crush me out.
00:33:32
>> But that's a good analogy because Google
00:33:33
is basically the other photographer and
00:33:35
Sony
00:33:36
>> and that's why we're we're like
00:33:39
>> they should know better.
00:33:40
>> We're just disappointed. I think in this
00:33:42
analogy, Google makes the Frisbee.
00:33:47
>> We're so deep down this rabbit hole.
00:33:49
>> It's called an ultraar actually.
00:33:51
>> Yeah. But I'm saying the tool is being
00:33:55
used differently.
00:33:57
>> Oh, so they made a bad product.
00:33:58
>> So they made a disc that flies slightly
00:34:00
differently. And us pro throwers are
00:34:03
like, "Ah, this turns way too fast or
00:34:06
this is like hanging in the air too
00:34:07
long." But regular people are like, "It
00:34:08
floats forever. This is great." Oh yeah,
00:34:11
>> that's that's good. So yeah, anyway,
00:34:13
that's good. Let the tools do the tool
00:34:15
thing. Okay, but there should always be
00:34:16
Pro Tools.
00:34:17
>> Okay,
00:34:18
>> I haven't experienced the new soundboard
00:34:20
yet.
00:34:20
>> It's pretty good.
00:34:21
>> Great.
00:34:21
>> Um, Gro contacts, please.
00:34:24
>> And with that note,
00:34:25
>> And with that note,
00:34:26
>> I think we should uh we should take a
00:34:28
quick break.
00:34:29
>> Let the steam out of our ears a little
00:34:31
bit. But before that,
00:34:32
>> subscribe and Google IO next. And before
00:34:34
that, and after that,
00:34:36
>> let's do trivia.
00:34:38
Yeah, this is going to be an episode for
00:34:40
sure.
00:34:42
>> It's already like 98 degrees in here.
00:34:45
>> I still can't remember when I said that.
00:34:47
>> Do it again. Can you do that again?
00:34:49
>> It's the waveform bros.
00:34:52
>> Why did I say it last?
00:34:53
>> It's funny that I know that that's
00:34:55
Marquez cuz I've heard that impression
00:34:56
before. But
00:34:57
>> but like was I responding to something
00:34:59
waveform bros?
00:35:00
>> Was I imitating something? I don't know
00:35:02
why I said that or said it like that.
00:35:04
>> He probably remember said it when we
00:35:06
weren't here. No, he I was here for
00:35:08
sure.
00:35:09
>> And we also have this one.
00:35:12
>> That could have been any of us.
00:35:14
>> Today's question is Rachel, Mimi, Robin,
00:35:19
and Shakira are all code names for what
00:35:23
line of smartphones?
00:35:24
>> Rachel, Robin, Mimi, Shakira.
00:35:27
>> Shakira.
00:35:29
>> Shakira.
00:35:30
>> Shakira.
00:35:30
>> Shakira. Shakira.
00:35:32
>> Shakira. Shakira. really know
00:35:37
>> the hips.
00:35:39
>> Interesting. Okay,
00:35:40
>> don't lie, Marz.
00:35:41
>> Well, think about it. I
00:35:42
>> Shakira is so good.
00:35:43
>> I have no idea about this.
00:35:44
>> The Zootopia songs are
00:35:46
>> What do you think about like Oh, this
00:35:47
this is like a smartphone whose code
00:35:49
name was Shakira. It must be really
00:35:50
groovy. Adam, look at this phone and
00:35:52
tell me how groovy it is.
00:35:53
>> Did you hear that um Brazil was trying
00:35:56
to like sue her for tax fraud?
00:35:57
>> No, it was Spain.
00:35:59
>> Spain was trying to sue her for tax
00:36:01
fraud. She got out of it and she well
00:36:03
she acquitted herself basically and was
00:36:05
like
00:36:06
>> did her hips go on the stand cuz they
00:36:07
can't lie.
00:36:11
>> Nice. All right, we really got to move
00:36:14
on.
00:36:15
>> D
00:36:19
me.
00:36:19
>> Yeah, that was you.
00:36:20
>> I don't remember that either.
00:36:21
>> We have a lot of good ones.
00:36:22
>> So many people have already tried to
00:36:24
skip the ad break and they're still
00:36:26
getting us yapping about things and it's
00:36:28
not even an ad yet. Okay, we'll think
00:36:29
about it. Anthony will be at the end
00:36:30
like usual.
00:36:31
>> The long way
00:36:34
>> and we'll be right back.
00:36:44
This podcast is supported by Chefman.
00:36:46
Quick Father's Day reminder. It's close.
00:36:48
And if you're still blanking on what to
00:36:50
get your dad, I get it.
00:36:51
>> Listen, dads are weirdly hard to shop
00:36:53
for. I know from experience. Um, we
00:36:56
either buy what we already want or we
00:36:58
say we want nothing. But as a newly
00:37:00
two-time father, here's the move. Ready?
00:37:02
Get something he'll actually use. It's a
00:37:04
good idea. We've both been actually
00:37:05
using the Chef IQ Sense and it's one of
00:37:07
those products that immediately earns a
00:37:09
permanent spot in the cooking drawer.
00:37:10
This is a wireless uh smart thermometer
00:37:13
that connects to your phone and
00:37:14
basically coaches you through cooking
00:37:15
meat perfectly every single time.
00:37:17
>> And we're not just saying that as a
00:37:19
slogan. I use it. It removes the most
00:37:21
annoying part of cooking meat, which is
00:37:24
guessing. I've always been a huge
00:37:25
proponent of we cook meat to
00:37:27
temperature, not to time. Um, so you put
00:37:30
the probe in, it tells you when to flip,
00:37:31
when to pull it, when to let it rest,
00:37:33
when it's ready to serve. It's the
00:37:35
difference between hovering over the
00:37:36
grill, stressing out, or just cooking
00:37:38
with confidence. Also, it's not just a
00:37:40
grill thing. I've used it on the grill,
00:37:41
but also in the oven, on the stove top,
00:37:43
and even with longer cooks. It works
00:37:45
with basically any protein and any
00:37:47
method. So, yeah, it's a great Father's
00:37:49
Day gift, and it's perfect because it's
00:37:50
actually useful. So, if your dad likes
00:37:52
cooking at all, this is the type of gift
00:37:53
that he's going to keep using, not just
00:37:55
a onetime gimmick.
00:37:56
>> So, we've partnered with Chef IQ for
00:37:58
Father's Day to get you 40% off. That's
00:38:00
a real discount and it's an easy win.
00:38:02
>> So, links down below. Use code wave for
00:38:05
40% off at chefiq.com. So, that's code
00:38:08
wave@ chefiq.com.
00:38:11
Get your dad the gift he'll actually
00:38:13
use.
00:38:15
Thanks.
00:38:17
All right, welcome back. This past week
00:38:19
was also Google IO. Now, we did talk
00:38:22
about the Android show the previous
00:38:23
week, which was like Google took out all
00:38:26
of the stuff that you might typically
00:38:28
expect of IO's in the past, which is
00:38:30
like Android updates, feature updates,
00:38:32
Android Auto, like the interesting
00:38:34
exciting stuff that's going to hit
00:38:35
consumers and that we can actually use
00:38:37
and review. They took that out and made
00:38:38
it a separate show a week before IO. We
00:38:40
talked all about that. So, that what
00:38:42
what does that leave for IO?
00:38:44
>> That leaves a lot of Gemini. Now we know
00:38:48
that IO is a developer conference and
00:38:51
they have to talk to developers and they
00:38:53
have to talk to the people on the
00:38:54
bleeding edge of like using these you
00:38:56
know tools essentially Gemini and all
00:38:59
the other things that they're we're
00:39:00
going to talk about. But uh I think as a
00:39:02
whole this IO since it was devoid of the
00:39:05
Android stuff felt like an overwhelming
00:39:08
amount of especially named
00:39:11
>> AI features. And if there's one thing
00:39:14
about Google, it's that they can't help
00:39:16
naming everything
00:39:18
>> everything
00:39:18
>> everything that they do. So, we're going
00:39:20
to jump into it. We can try to explain a
00:39:22
lot of this new stuff, how much of it
00:39:24
you may end up using someday or how much
00:39:26
of it is really technically advanced or
00:39:27
cool, but I promise you, you will not be
00:39:30
able to keep track of how many names
00:39:31
there are.
00:39:32
>> Yeah. I think that this IO was like the
00:39:34
biggest example yet of the fact that
00:39:37
Google is like 3,000 companies in one.
00:39:40
>> Yeah. Yeah. And that they just don't
00:39:41
really communicate very much cuz there's
00:39:43
no cohesiveness or vision. It's just
00:39:45
spamming. Like it really does feel like
00:39:47
there's nine different companies who
00:39:48
each had three ideas to present and none
00:39:50
of them talk to each other. And
00:39:51
honestly, some of them step on each
00:39:53
other. Totally.
00:39:53
>> Like there's there's VO and then there's
00:39:55
also Nana Banana and then there's also
00:39:57
this new thing that's going to generate
00:39:58
images and videos. Like why are there
00:40:00
three? So there's a real lack of visual
00:40:03
strategy to what's happening.
00:40:05
>> So we'll break it down. We'll show you
00:40:06
what's new. But again, I promise
00:40:08
>> Google Photos. It's not just Google
00:40:10
Pics.
00:40:11
>> Google Pics. Brand new product.
00:40:13
>> It's not Google Photos or pictures.
00:40:15
>> Yeah.
00:40:16
>> But it will generate some of those
00:40:17
things.
00:40:18
>> So, I last night thoroughly watched this
00:40:21
like three times um to my own dismay.
00:40:23
>> Oh my goodness.
00:40:24
>> I'm so sorry. In 4x though, probably.
00:40:25
>> Uh in 2x.
00:40:26
>> Okay. It's a little less bad.
00:40:28
>> I tried to do 4x and I just couldn't
00:40:30
keep up.
00:40:30
>> There's still three hours if you watched
00:40:31
it three times.
00:40:33
>> Yeah. the the amount of times I gave the
00:40:36
show where they said like I think they
00:40:37
said and we're just getting started at
00:40:39
an hour and I was like what
00:40:41
>> don't do this to me. Yeah. Well, the
00:40:43
first time to be fair I listened to it
00:40:44
in the car at 2x and then when I got
00:40:45
home I watched it again. Anyway, for
00:40:47
some reason they bounced back and forth
00:40:50
between different products through the
00:40:52
show like they would announce something
00:40:54
like Omni and then later in the show
00:40:55
they'd talk about Omni again and then
00:40:57
near the end they'd talk about Omni
00:40:58
again. And it felt like because they
00:41:00
took everything out from the Android
00:41:02
show, they had to like pad the time
00:41:03
because Google IO is always two hours.
00:41:05
It's weird. I try to like think about
00:41:06
who's the target demographic of these
00:41:07
presentations. And maybe it's unfair to
00:41:09
compare it to Apple, but like I think
00:41:11
regular people watch a WWDC recap to see
00:41:14
what interesting things are going to be
00:41:15
happening.
00:41:16
>> Bunch of my friends watch Dubdub. I
00:41:18
don't even I'm not even going to make an
00:41:20
IO recap because so much of this is
00:41:22
either not going to come out or it's
00:41:23
going to be dead in a year or it's going
00:41:25
to be really interesting and then it's
00:41:26
going to get stepped on by some other
00:41:28
updated Google product in like 3 months.
00:41:30
>> Yeah.
00:41:30
>> So yeah, there's a lot of very
00:41:32
technically advanced stuff and
00:41:34
impressive things that they got to demo.
00:41:36
>> How much of it will actually show up?
00:41:38
Like the live docs feature was really
00:41:40
cool. Are they going to ship it? I don't
00:41:41
know. Is it going to last? I don't know.
00:41:43
But it was a really cool demo. So we'll
00:41:45
tell you about it. I was also very
00:41:47
confused about who this was for because
00:41:49
IOIO was supposed to be for developers
00:41:52
>> and I feel like a lot of this wasn't
00:41:53
even aimed at developers. It was like
00:41:55
them showing off their dominance in web
00:41:57
and search
00:41:58
>> and token usage.
00:41:59
>> Yeah.
00:41:59
>> Yeah. That was a weird flex multiple
00:42:02
>> Sundar was just flexing constantly about
00:42:03
how many tokens they consume and then at
00:42:05
some point he was just like a salesman
00:42:07
and he was like if your company used
00:42:09
Gemini Flash 4.5
00:42:12
billion
00:42:12
>> you would save a billion dollars. did
00:42:15
his one stat was like
00:42:17
>> 375 people have used one trillion
00:42:21
tokens.
00:42:22
>> Yeah.
00:42:22
>> Like that which
00:42:24
>> is such a weird
00:42:25
>> also weird because flash 3.5 is more
00:42:28
expensive
00:42:29
>> than three was. I think some of these
00:42:31
>> 3.1 so it's like
00:42:33
>> some of these things just felt like them
00:42:34
saying hey we are the dominant AI
00:42:37
company.
00:42:38
>> Yeah.
00:42:38
>> See these big numbers.
00:42:40
>> Yeah.
00:42:40
>> That's how you know. I will say it's
00:42:42
interesting that it's a developer
00:42:44
conference, but in the past years
00:42:46
they've had enough so that
00:42:47
non-developers like people like us, like
00:42:50
the media
00:42:51
>> can like really understand and digest
00:42:52
and understand like how you can use this
00:42:54
in the real world.
00:42:56
>> And this year it was it didn't feel like
00:42:59
it was for developers really, but it
00:43:01
because it was just products that were
00:43:02
being announced, but none of the
00:43:03
products were things that people would
00:43:05
actually use. So, only a few of them. Or
00:43:08
if you are like an enthu like a real
00:43:09
enthusiast who's like, I can't wait to
00:43:11
deploy Gemini Spark to shop for me. Like
00:43:13
there's there's a tiny fraction of
00:43:15
people who are like, "Hell yeah, this is
00:43:16
awesome. I can't wait for it to monitor
00:43:18
the price of jeans online to
00:43:20
automatically buy six pairs for me when
00:43:22
it hits a certain price threshold." But
00:43:23
like that's not most of us. I remember
00:43:25
like 8 years ago at IO, we'd all sit
00:43:27
around being like, "What are they going
00:43:28
to name the next Android?" Yeah.
00:43:30
>> And what's the big feature or two going
00:43:32
to be that they try to like show us is
00:43:33
awesome? What's the new Nexus phone that
00:43:35
is going to be under our seat when we
00:43:37
the show finishes?
00:43:38
>> When are we going to delete that chain
00:43:39
link fence in front of this?
00:43:41
>> Never.
00:43:43
>> Yeah. So, because they were hopping
00:43:44
around so much throughout the show, I
00:43:46
had originally just taken a ton of notes
00:43:48
in like a linear fashion and then I
00:43:51
decided to order them into main chunks.
00:43:53
So, I have a models section.
00:43:55
>> That's where you messed up. You should
00:43:56
have just asked Gemini to do it for you.
00:43:58
>> Um, I did.
00:44:00
>> Which one?
00:44:02
I I did have Jeff and I do this for me,
00:44:04
but then I went through and I like Yeah,
00:44:06
I Yeah. Anyway,
00:44:08
>> okay. So, we have a model section, we
00:44:10
have an agentic AI section, we have a
00:44:12
retail section.
00:44:13
>> Okay.
00:44:13
>> We have an intelligent eyewear section
00:44:15
which is like the glasses
00:44:16
>> app and search overhaul section. And
00:44:18
then the little ending weird deep mind
00:44:20
thing um where we're on the foothills of
00:44:22
the singularity apparently. Should we
00:44:24
also save the intelligent eyewear
00:44:25
section for the end for attention like
00:44:27
Google did?
00:44:28
>> Sure, we could do that.
00:44:29
>> They got to that like 90 minutes into
00:44:31
IO. Yeah, which is crazy.
00:44:32
>> Yeah. All right. Well, we'll start with
00:44:34
the models. Uh, new models obviously
00:44:36
Gemini 3.5 Flash and there will be a 3.5
00:44:40
Pro coming in a month apparently. Um, I
00:44:42
don't know why it's taken so long to
00:44:43
roll that out. That's kind of weird.
00:44:45
Google claims that Flash is four times
00:44:47
faster in out output tokens per second
00:44:49
than existing Frontier models and
00:44:51
outperforms Gemini 3.1 Pro on key
00:44:54
benchmarks, which is impressive if
00:44:56
that's true.
00:44:57
>> I want to be like the regular person
00:44:59
translator for these. Yeah, do it.
00:45:01
>> I think that that essentially would mean
00:45:03
for the super quick queries that you're
00:45:05
going to IO to AI for that use flash,
00:45:09
they will be really really good because
00:45:11
they're benchmarking as well as pro and
00:45:12
they'll be really really fast. So you
00:45:14
don't have to wait for it to type it out
00:45:16
in several seconds of is AI going to
00:45:17
even get this right? Ideally it's way
00:45:19
faster and way more accurate.
00:45:21
>> Yeah, ideally.
00:45:22
>> Yeah. To me, it also felt like Sundar
00:45:24
was trying to get Enterprise to sign up
00:45:26
to use Gemini as their back end because
00:45:28
Anthropic is currently making a load of
00:45:31
money on that.
00:45:32
>> Yeah. Hey, do you make an app? Do you
00:45:34
have like a thing that queries AI inside
00:45:36
your app? Well, make sure it's Gemini
00:45:38
Flash because it'll be the fastest, best
00:45:40
one, best experience for your app user.
00:45:42
>> Yeah,
00:45:43
>> sure.
00:45:43
>> Yeah.
00:45:44
>> Uh rolled out already.
00:45:46
>> Yeah, it's already out. That's true.
00:45:47
Yeah.
00:45:47
>> Yeah. Uh they also demoed the Gemini
00:45:49
Omni World model. So, this one kind of
00:45:53
confused me a little bit. Um, what it is
00:45:56
good at specifically is simulated
00:45:57
physics in real world reasoning to
00:45:59
interpret combinations of text, audio,
00:46:01
images, and video to generate realistic
00:46:03
editable video environments. Uh, they
00:46:06
gloated a lot about how the model can do
00:46:08
everything, but said, "For now, it's
00:46:10
only doing video." And I'm like, "Okay,
00:46:14
it can do everything, but it can only do
00:46:16
video."
00:46:17
>> Yeah. So my translation of this is like,
00:46:20
hey, have you ever wanted to make an AI
00:46:22
generated video, but you wanted to use
00:46:24
some inspiration from a photo and a
00:46:27
video and some text that you wanted to
00:46:30
describe some extra stuff in it and some
00:46:33
other random thing you wanted to
00:46:34
include. Well, you can now just loop all
00:46:36
of that stuff in. It'll be one prompt
00:46:38
and Omni will turn it into something
00:46:40
that includes all of that stuff. Which
00:46:42
the irony of all this is that I think
00:46:44
the hierarchy of usability and value in
00:46:47
AI is like coding is number one, text
00:46:50
generation is probably number two. But
00:46:52
people like AI less and less the more
00:46:54
you get towards video, right? Like AI
00:46:57
image generation mostly bad, but you
00:47:00
could use it for manipulation to make
00:47:02
memes of yourself for your
00:47:03
>> rage bait. Misinformation is making a
00:47:05
ton of money online. This is perfecto.
00:47:08
>> And then but then video like I don't
00:47:10
really know anybody who is hyped for AI
00:47:12
video in any possible way.
00:47:14
>> Yeah.
00:47:14
>> In general,
00:47:14
>> I could even like think about a couple
00:47:17
things. I could use AI generated images
00:47:19
for like visualizing like, hey, I've got
00:47:21
this room. I want to visualize what this
00:47:24
desk would look like in here with this
00:47:25
wallpaper behind it. Okay, AI can help
00:47:27
me see what that looks like.
00:47:28
>> Right.
00:47:29
>> The video brainstorming maybe. I don't
00:47:32
feel like that's going to be very
00:47:33
useful.
00:47:34
>> Yeah. And they really did a lot of AI
00:47:37
videos.
00:47:37
>> And it's the heaviest resource-wise.
00:47:40
Like it's the most difficult to do.
00:47:41
>> Yeah.
00:47:42
>> And it's Yeah. Yeah.
00:47:43
>> Will Smith eating spaghetti.
00:47:45
>> Yeah.
00:47:45
>> It's getting better.
00:47:46
>> And they specifically were like, it's
00:47:48
it's multimodal. It can do everything.
00:47:50
And I'm like, well, Gemini was already
00:47:52
multimodal, so why do you need to call
00:47:53
it Gemini Omni? Why didn't they just
00:47:55
make an update to Gemini and say we can
00:47:57
now do real world physics?
00:47:58
>> Because they need to name everything.
00:48:00
>> This is what we talked about.
00:48:02
>> Yeah. New Gemini Omni. The naming team
00:48:04
needed a raisin. They went all out for
00:48:06
this.
00:48:06
>> Yeah. Okay. So, that's it for the
00:48:08
models. Uh there's new Agentic AI
00:48:10
updates. We have anti-gravity 2.0, which
00:48:13
I know Adam was actually a little bit
00:48:14
excited for. If you don't remember what
00:48:15
anti-gravity was, it's basically their
00:48:17
coding environment, uh that has all
00:48:20
these agentic features where you can,
00:48:21
you know, code inside of it and stuff.
00:48:23
Um
00:48:24
>> yeah, a lot of people were confused
00:48:25
because anti-gravity 2.0 is very much
00:48:29
like the claw desktop app basically or
00:48:31
like Codex, I've heard. I haven't used
00:48:32
codeex but similar to that where it's
00:48:34
just like a chat or whatever. Yeah, they
00:48:35
still have the IDE.
00:48:37
>> So anti-gravity was their IDE but I
00:48:41
don't know what got lost in translation.
00:48:43
I don't know why they named it the same
00:48:44
thing. So this new thing now is
00:48:46
anti-gravity that you talk to Gemini and
00:48:49
it'll like do the things. But if you
00:48:50
want to open up the code and look at it
00:48:52
with a real proper file structure and
00:48:54
everything, you need to also download
00:48:56
the anti-gravity IDE which is a
00:48:58
completely separate thing. And I don't
00:48:59
know how much longer that's going to be
00:49:00
around. I don't I don't know. It's very
00:49:02
confusing.
00:49:03
>> It looked I know. It looked a lot like a
00:49:05
general like a regular Gemini app.
00:49:07
>> Like I
00:49:09
>> It's strange. Anyway, uh it's for
00:49:11
developers.
00:49:12
>> Wait, can I just Sorry.
00:49:13
>> Yeah, go for it.
00:49:14
>> So, if you're using anti-gravity 2.0 to
00:49:17
make something, but you want to make
00:49:18
sure you want to look deeper, you have
00:49:20
to open up anti-gravity.
00:49:22
>> Well, so anti-gravity, if you if you
00:49:26
tell it to build the thing, it will do
00:49:27
it. And then it'll be like you can click
00:49:29
here to look at the code and it'll pop
00:49:30
up a little side window and preview like
00:49:32
the actual code. You can scroll through
00:49:33
it, whatever. But that's not typically
00:49:35
how people build things. They have an
00:49:37
editor for that. So anti-gravity I
00:49:40
believe was it's very similar to VS
00:49:42
Code, which is a very popular one open
00:49:44
like whatever, but that's a separate app
00:49:47
entirely now on on
00:49:49
>> the computer.
00:49:50
>> I think if you want to know if the
00:49:52
Google teams talk to each other, there's
00:49:53
your proof that they don't talk to each
00:49:55
It did feel to me like there's one tool
00:49:57
for generating code with AI and then
00:50:00
there's another tool with a different
00:50:01
name for auditing your AI generated code
00:50:04
with another AI.
00:50:06
>> But it sounds like they're the same
00:50:07
name.
00:50:07
>> But if Yeah. And then also like your
00:50:09
regular code. So you can audit all of
00:50:11
your code with AI.
00:50:12
>> AI coding sometimes makes mistakes. So
00:50:14
you need someone or something to audit
00:50:17
mistakes so another agent can audit your
00:50:19
code. And if that's not confusing
00:50:20
enough, there's also the anti-gravity
00:50:22
CLI now, which is another terminal tool
00:50:26
that is replacing the Gemini CLI, which
00:50:28
was open source, and now it's not open
00:50:31
source. So now you can control all of
00:50:33
your agents all in one in the terminal,
00:50:35
but you have to use their their
00:50:37
proprietary.
00:50:37
>> That's also called anti-gravity.
00:50:39
>> And that's also called anti-ra the
00:50:40
anti-gravity CLI.
00:50:41
>> Jesus. Yeah.
00:50:42
>> Yeah. It's it's going great.
00:50:43
>> Even if I was a developer, I would be
00:50:45
really confused. Do you think anyone's
00:50:47
showed up to their first day at work at
00:50:48
Google and they're like, "I'm on the
00:50:49
anti-gravity team and they went to the
00:50:51
wrong building." It's like in school
00:50:52
when you sit in the wrong classroom and
00:50:54
20 minutes in they're reading the
00:50:55
syllabus and you're like,
00:50:56
>> "Do I get up and leave or do I
00:50:58
>> before for sure?"
00:50:59
>> It's like I get that they all do
00:51:01
slightly different things, but they
00:51:02
should be one thing.
00:51:04
>> Should be cohesive.
00:51:06
>> Well, and the the irony of all of this
00:51:08
is the general Gemini app can basically
00:51:10
do all of this stuff, too. Like I've
00:51:12
even like used the Claude app on on my
00:51:14
Mac and I've like updated my website
00:51:16
with features and stuff through Claude
00:51:19
and you can just do it through the chat.
00:51:21
You don't even have to use clog code,
00:51:23
but there is a cloud code. Yeah.
00:51:25
>> So it's it's like they make all of these
00:51:27
sub products that are more optimized for
00:51:30
the main thing, but the main model can
00:51:32
still do all of the things
00:51:34
>> probably to like some 90s something
00:51:35
percentile
00:51:36
>> of the specialized tool.
00:51:38
>> Yeah.
00:51:39
>> Yeah.
00:51:39
>> Yeah. I don't know. Well, they were very
00:51:41
excited about anti-gravity 2.0. They
00:51:43
built an operating system from scratch.
00:51:45
>> Yeah.
00:51:46
>> Plugged it into everything. Like they
00:51:48
were showing Google search things that
00:51:50
were building stuff with anti-gravity.
00:51:52
All this stuff
00:51:52
>> which we'll we will get to. Um but yeah,
00:51:55
they built this operating system. They
00:51:56
said it deployed 92 sub aents over 12
00:51:59
hours to build an OS and that it cost
00:52:03
less than $1,000 in API credits.
00:52:06
>> Wow.
00:52:07
>> Um
00:52:08
what use this has? who needs to build
00:52:10
operating systems. I'm not sure, but I
00:52:13
guess it was more of a the scientists
00:52:15
were more interested if they could, not
00:52:16
if they should situation.
00:52:18
>> Yeah. Uh and then they ran Doom on it.
00:52:20
>> They could run Doom. Yay.
00:52:21
>> So,
00:52:22
>> look, we're relatable.
00:52:23
>> Yeah. Relatable. Okay. Gemini Spark. Uh
00:52:26
this is basically Google's answer to
00:52:27
OpenClaw. It's a 247 cloud-based
00:52:30
personal AI agent that runs continuously
00:52:32
in the background. Um, it can actively
00:52:35
parse your data to give you daily
00:52:36
digests like a child's school emails for
00:52:39
deadlines, monitoring your credit card
00:52:40
statements, etc. Um, it's rolling out to
00:52:43
trusted testers this week and rolling
00:52:45
out to a new $100 AI ultra plan. So,
00:52:48
this is the other thing that they have
00:52:49
multiple the same name for multiple
00:52:51
things. They now have three AI Ultra
00:52:54
plans
00:52:54
>> at three different prices
00:52:55
>> at three different prices that are not
00:52:57
called like AI Ultra Plus or AI Ultra
00:53:00
Ultra. They're just AI ultra, but
00:53:02
there's the $100 one, there's the two
00:53:04
now $200 one down from 250,
00:53:07
>> and then they have the crazy one, I
00:53:09
think. Um, unless it's just the $200.
00:53:13
Anyway,
00:53:13
>> so the way I'm thinking about this one
00:53:15
for regular people,
00:53:16
>> the open claw thing,
00:53:17
>> the Yeah, spark. Like, if you think
00:53:19
about using AI as I do prompt, then AI
00:53:23
does thing, then I do prompt again, then
00:53:25
AI does adjustment, then I prompt again,
00:53:27
and it's like a back and forth thing.
00:53:28
>> Yeah. The idea of an agent is it's able
00:53:31
to go and do multiple things. It has its
00:53:33
own agency. It has like the ability to
00:53:35
do multiple things over time or in a row
00:53:38
or to go execute something for you
00:53:40
>> without you having to prompt over and
00:53:42
over again. Right?
00:53:43
>> I did find this decently useful and like
00:53:46
and this is an example that'll come up
00:53:47
again of like I want to monitor when
00:53:49
this thing drops below a certain price,
00:53:52
but you'll have to check the market for
00:53:53
listings of it over and over again.
00:53:56
Let's do this every 12 hours for the
00:53:58
next couple weeks until we find one that
00:53:59
hits this certain price.
00:54:00
>> Yeah,
00:54:01
>> that that's what the agent would do. It
00:54:02
would just go, "Hey, I searched again.
00:54:04
Here's the results. Hey, I searched
00:54:05
again. Here's the one that actually hit
00:54:06
your price threshold." Yeah.
00:54:08
>> So, it's instead of you searching over
00:54:09
and over through the agent, it does it
00:54:11
>> with agency.
00:54:12
>> Yeah. Yeah. And like Spark,
00:54:14
>> all of these AI companies have come up
00:54:16
with sort of an open claw competitor.
00:54:18
Like Claude now has one that can do it
00:54:20
for you. Mhm.
00:54:21
>> Um, one of my friends, they lost a bunch
00:54:23
of data on their hard drive. And so they
00:54:24
were using a recovery tool, but the
00:54:26
recovery tool estimated like 800 hours
00:54:28
to like recover the data. So he set it
00:54:31
up so that the agent would check his
00:54:32
computer and check on the progress of it
00:54:34
like every once like one day and send
00:54:36
him an update.
00:54:37
>> So there are like some uses there. Um,
00:54:40
>> I will say the daily digest I'm actually
00:54:42
pretty excited about because I've been
00:54:43
getting like the beta one that gets like
00:54:45
emailed to you in the morning.
00:54:46
>> So bad.
00:54:47
>> Really? I found it pretty helpful. I
00:54:49
like the formatting of it, but the
00:54:51
contents of it are useless to me.
00:54:52
>> Oh, mine is the opposite. I hate the
00:54:54
formatting of it, but I've actually
00:54:56
found a couple subscriptions that were
00:54:58
like, "Oh, yeah, I do have to cancel
00:54:59
that free trial." You know, like it
00:55:01
reminds you of things like that.
00:55:02
>> I think potentially I've connected it to
00:55:03
too many services cuz it'll just be
00:55:05
filled up with like a bunch of people
00:55:07
left comments on your Google Docs and
00:55:09
I'm like, I already know that. Like, I
00:55:10
haven't marked them as resolved cuz I
00:55:12
need to read them later. And so, it's
00:55:13
just telling me a bunch of things I
00:55:14
already knew. It's like, here's what's
00:55:15
on your calendar. Well, yeah, I got a
00:55:17
notification. So, I have the calendar
00:55:18
why the calendar notifies me.
00:55:20
>> But yeah, maybe I can fine-tune it a
00:55:22
little bit.
00:55:22
>> Yeah,
00:55:22
>> it's fine. It's totally worth a trillion
00:55:25
dollars of investment.
00:55:26
>> Why don't you get an agent to finetune
00:55:27
it?
00:55:29
>> I'll deploy Gemini Spark, deploy 92 to
00:55:32
finetune it.
00:55:33
>> Uh that's coming to the Mac desktop app
00:55:35
later this summer for local file
00:55:36
automation. So, it'll be able to like
00:55:38
actually touch your computer because
00:55:40
right now it can only do things like on
00:55:41
the web. So,
00:55:43
>> okay. Um three is retail. This is a
00:55:46
grouped a grouped thing for universal
00:55:48
cart and AP2. So this universal cart
00:55:49
thing is is kind of interesting. Clearly
00:55:52
Google has been trying to make Google
00:55:54
shopping happen for like 12 years now,
00:55:57
15 years now where they just they really
00:56:00
really see all the money that Amazon's
00:56:02
making and they're like we could have
00:56:03
some of that.
00:56:04
>> They want to be a trusted retailer so
00:56:05
bad.
00:56:06
>> So bad. So and then they saw what Tik
00:56:08
Tok was doing with Tik Tok shop. up. And
00:56:10
so now if you look at YouTube and you go
00:56:11
to like, you know, anybody's YouTube
00:56:13
video that has a a product placement,
00:56:16
they now have right below the video like
00:56:18
you can buy this and you can click on
00:56:19
it, you can go to it and buy it,
00:56:21
whatever. Probably not enough people use
00:56:23
that that they want to use it. So they
00:56:25
created this thing called universal
00:56:27
cart. Um, which basically across Google
00:56:30
search, Gemini, YouTube, and Gmail. You
00:56:33
can add things to your cart. you can
00:56:36
click on a thing and say you're
00:56:37
interested in it, but when it drops
00:56:39
below a certain price, you're kind of
00:56:40
interested in it and it connects to your
00:56:42
Google wallet. So now as well, if it
00:56:46
knows like that you're you have a
00:56:47
certain credit card and that certain
00:56:48
credit card gets certain cash back at
00:56:50
certain stores,
00:56:52
>> it will be like, "Oh, when you go to pay
00:56:54
for this, you should use your Target
00:56:55
card because it's at Target and you get
00:56:57
5% off at Target or something like that.
00:56:59
You get cash back, whatever."
00:57:01
Um, I really really think Google is just
00:57:03
trying to find more revenue sources
00:57:05
because by making Gemini and making all
00:57:07
this automation stuff, they are taking
00:57:09
away from search ads. So, they just
00:57:12
they're slowly trying to they're like,
00:57:14
"We're going to get ahead in the AI
00:57:16
thing, even though it's taking away from
00:57:17
our main business, but we need to figure
00:57:19
out ways to make money from it if we're
00:57:22
going to own it."
00:57:23
>> Yeah.
00:57:23
>> Yeah. So, it's kind of interesting, but
00:57:27
I remember seeing the study where there
00:57:29
was a bunch of uh they tested a bunch of
00:57:31
these different models at how good is it
00:57:35
at actually suggesting me like the right
00:57:36
flights, the right hotel reservations,
00:57:38
whatever. If you were just to say like
00:57:40
make me this trip
00:57:42
>> classic.
00:57:42
>> And like 90% of the time it gave the
00:57:44
users like one of the more expensive
00:57:47
options for the hotel, for the flight,
00:57:48
all these things.
00:57:49
>> So classic. And so Google can now go to
00:57:52
you know retailers and advertisers and
00:57:54
they can say hey look we can slow slowly
00:57:57
sort of like suggest these products
00:58:00
towards these users
00:58:02
>> and kind of the overall the overarching
00:58:04
theme for IO this year is just like
00:58:07
trust us more and more and more bro
00:58:09
>> which makes me trust you less and less
00:58:11
and less.
00:58:11
>> Yeah.
00:58:12
>> Yeah. Because all I see with all of this
00:58:14
is I think you guys mentioned it on an
00:58:15
episode I wasn't here is like the new
00:58:17
version of SEO because all of this is
00:58:19
just going to lead into a different way
00:58:21
that big companies can pay to be better
00:58:23
results inside of AI.
00:58:24
>> Exactly.
00:58:25
>> That is 1,000% where it's going. No
00:58:27
matter what you say, that's how it's
00:58:29
going to work.
00:58:30
>> Yeah.
00:58:30
>> That's how people are going to make
00:58:31
money on it.
00:58:31
>> Google is just like, "Oh, just don't
00:58:33
make your own decisions. Like, let AI
00:58:35
make all the decisions for you." And
00:58:36
then they're going to advertisers and
00:58:37
they're saying, "Now that everyone just
00:58:39
trusts us, we can just suggest your
00:58:41
products." Yeah, they don't even
00:58:43
>> It points pretty directly to the direct
00:58:45
trade-off between privacy and
00:58:46
convenience because and this is also a
00:58:49
shout out to Joanna's book, which we'll
00:58:51
you'll you'll hear about that probably
00:58:52
on an upcoming bonus episode, but like
00:58:54
if you just give your life to the AI,
00:58:57
you just go, "All right, you know
00:58:58
everything about me, Google. It's kind
00:59:00
of a privacy nightmare, but you have
00:59:01
everything about me now." That is the
00:59:04
best case scenario for actually getting
00:59:06
the results that are most applicable to
00:59:08
you. And so when you ask for it, yes, it
00:59:10
will be skewed, but it will be skewed
00:59:12
based on what it knows about you. And
00:59:14
theoretically, it'll give you better
00:59:15
products to buy or more useful
00:59:18
suggestions based on it knows about your
00:59:20
car, your house, or whatever already.
00:59:22
>> And so if you do decide to just give it
00:59:24
all to the AI, then that will be your
00:59:26
win. But not everybody wants to do that.
00:59:29
And obviously that's the trade-off that
00:59:30
they're trying to convince you to make.
00:59:32
It's why a lot of people like keep the
00:59:34
Instagram ad tracking on because they'll
00:59:36
say like, "Oh, I actually like it
00:59:37
because it gives me relevant products
00:59:39
and I'd rather have relevant products
00:59:41
than, you know,
00:59:42
>> turn it off."
00:59:43
>> The lingerie that I get.
00:59:45
>> There's literally no reason to have that
00:59:46
on.
00:59:47
>> Wrong. I've gotten so many good
00:59:49
products.
00:59:50
>> People want to get I told you there's a
00:59:52
divide.
00:59:53
>> Some people actually do want to buy
00:59:54
stuff from Instagram.
00:59:54
>> It's not that I want to buy stuff from
00:59:56
Instagram. It's that it shows me brands
00:59:58
from smaller brands like that aren't
01:00:00
just like Amazon or Sheen, you know?
01:00:01
Like I find actual stores from Brooklyn
01:00:03
or something cuz they're using targeted
01:00:05
ads. It's good for small businesses.
01:00:07
It's bad for me, but it's good for small
01:00:09
businesses.
01:00:10
>> That's That's Facebook.
01:00:12
>> Bad argument. Bad take. Bad argument. Do
01:00:14
not turn on app tracking for the sake of
01:00:17
helping small businesses. That is
01:00:18
ridiculous.
01:00:19
>> That was turn helping small businesses
01:00:22
by by letting one of the largest
01:00:24
companies in the world sell your data.
01:00:25
Are you serious, bro?
01:00:26
>> How else are you going to find The small
01:00:27
business has been spending money
01:00:29
desperately trying to find good
01:00:30
customers and this is the way they
01:00:33
actually
01:00:33
>> No, I'm not I'm not accepting any of
01:00:35
these.
01:00:35
>> I only make money
01:00:38
jersey. I sold it.
01:00:40
>> I know a few people who like use meta
01:00:42
advertising and they're literally like
01:00:44
you put in $10 and $30 pops out.
01:00:47
>> So clearly there it works.
01:00:50
>> There's a lot of wrong with Meta,
01:00:51
but that product is a money fountain.
01:00:53
There's a reason Meta is a giant giant
01:00:56
giant corporation and that small like
01:00:58
everyone uses it. I don't think that you
01:01:00
know only small business whatever but if
01:01:01
you are a small business and you put $10
01:01:03
in and $30 pops out. I mean
01:01:06
>> unrelated and related they're doing a
01:01:09
ton of layoffs today. So clearly this is
01:01:11
not great timing for any of this meta
01:01:14
stuff but they laid off like 10,000
01:01:16
people this week
01:01:17
>> again.
01:01:17
>> Yeah.
01:01:18
>> 8,000. Yeah.
01:01:19
>> Oh my god. Okay. Well, that's great. Uh
01:01:22
AI. Okay. The other thing in retail is
01:01:24
there is now this thing called AP2 which
01:01:27
is agent payments protocol
01:01:29
>> and uh basically Google is scared that
01:01:31
the agents are going to hallucinate with
01:01:33
your credit card information. That's
01:01:35
what they are terrified about there
01:01:36
because they you know they they want you
01:01:38
to be able to tell your agent like oh if
01:01:41
this brand like if this shirt from this
01:01:43
brand goes on sale can you just buy it
01:01:44
for me? Mhm.
01:01:45
>> Which is kind of crazy to think about
01:01:48
like
01:01:49
>> that an something could just agentically
01:01:51
like shop shop for you without you
01:01:53
having like discreet input. It's kind of
01:01:56
close to what someone so like an ultra
01:01:58
rich person who has like a personal
01:01:59
assistant.
01:02:00
>> Yeah.
01:02:00
>> They'd be like
01:02:01
>> I mean you'd be too rich to do this but
01:02:03
you'd be like when it hits a certain
01:02:04
price
01:02:05
>> just buy it for me.
01:02:06
>> Right.
01:02:07
>> But we don't have personal assistants.
01:02:09
But it's like oh yeah now my phone is a
01:02:10
personal assistant. I'll just have it
01:02:12
wait on something for me and it has
01:02:14
infinite capability of waiting and
01:02:16
finding the right thing and then it just
01:02:17
does it.
01:02:18
>> It's the weird thing of AI where it's
01:02:19
like that is a totally reasonable thing
01:02:21
to ask but at the same time weird robot
01:02:25
internet has my credit card number like
01:02:27
I like both of these things are true and
01:02:29
I'm terrified of
01:02:30
>> well it's kind of the next do you
01:02:32
remember Amazon Dash buttons?
01:02:33
>> Yes.
01:02:33
>> Yeah.
01:02:34
>> Yeah. So, for the young the youngans out
01:02:36
there, Amazon used to like give you for
01:02:40
$5 you could buy this Amazon Dash
01:02:41
button, then they give you a $5 credit
01:02:43
and it was a button that you could
01:02:44
assign to any product on Amazon. So,
01:02:46
when you hit it, it ordered it for you.
01:02:48
>> It was like if you put it by where you
01:02:50
keep your paper towels, like something
01:02:51
that you restock a lot on. So, like
01:02:53
every time my paper towels are almost
01:02:54
up, I just press it. They're going to
01:02:56
come to my door.
01:02:56
>> Tide pods. Tide pods. So, like what
01:02:59
Samsung and all these companies have
01:03:01
wanted to do forever is put cameras in
01:03:02
your fridge so it automatically does
01:03:04
this for you, right? That's like cuz
01:03:06
these companies, they're like, "Oh, if
01:03:07
people bought things as soon as they ran
01:03:09
out of them every single time and they
01:03:10
bought the same thing over and over and
01:03:12
over again, we could make more money
01:03:13
because it's more purchases per hour,
01:03:15
>> right? So, that's what Google's trying
01:03:17
to do here. They're trying to make it
01:03:18
like, oh, you're because all these ad
01:03:20
companies, the main problem that they
01:03:22
have with like the Facebook pixel, all
01:03:23
this stuff. They they see that you're
01:03:26
interested in something because you
01:03:27
interacted with it, but actually getting
01:03:30
you to make that payment and actually go
01:03:32
through with the purchase is the hardest
01:03:34
part of the whole thing. Yeah.
01:03:36
>> And so Google's like, "Well, what would
01:03:38
it take for you to buy this? Would it
01:03:40
take as frictionless as possible?
01:03:41
>> Would it take a $20 discount? What if it
01:03:43
got to a $20 discount? Would you
01:03:45
definitely buy it? Can I just do it for
01:03:46
you?" Like,
01:03:48
>> yeah. So, that's
01:03:49
>> that's a really good point. I feel like
01:03:50
there's so many times where I have
01:03:51
something in my cart and I almost buy it
01:03:52
and it's like,
01:03:53
>> no, not right now. But the the there's
01:03:56
way less commitment of being like, well,
01:03:58
you know, like if you found this thing
01:04:00
later and then it's just going to show
01:04:02
up at my door. Like, chances are if
01:04:04
that's what I was thinking in my head, I
01:04:05
would go back to that in two weeks,
01:04:07
probably get to the same point, be like,
01:04:08
I don't really need this.
01:04:09
>> Yeah, exactly.
01:04:10
>> Google's trying to get past the you
01:04:12
don't really need this. And that's
01:04:13
something that Meta doesn't have cuz the
01:04:15
the only thing Meta really has is when
01:04:16
you like hover over something for too
01:04:18
long, it'll just keep showing you the ad
01:04:20
like more and more often, you know?
01:04:22
>> Uh anyway, this agent pay payments
01:04:24
protocol, it's basically just a protocol
01:04:26
that's helping agents shop for you, but
01:04:29
making sure that they don't pass any
01:04:31
weird boundaries and buy things that you
01:04:33
don't want. Um, I think this is kind of
01:04:35
just security for them to get
01:04:38
advertisers to actually advertise to
01:04:40
with them and for them to be like,
01:04:42
"Don't worry, we won't make users return
01:04:45
a ton of stuff."
01:04:46
>> This has the potential to really screw
01:04:48
up some smaller businesses who might
01:04:50
have a bunch of accidental purchases
01:04:52
that then have to be returned and maybe
01:04:54
can't be put back on shelves or have
01:04:55
like
01:04:56
>> Yeah.
01:04:57
lots of accidental accidental purchases
01:05:00
that could create waste and kind of
01:05:01
screw over a small warehouse.
01:05:03
>> Yeah, because if you have returns and
01:05:05
then it's worth nothing and yeah, so
01:05:07
it's a whole thing.
01:05:08
>> All right, number four. App and search
01:05:09
overhauls. Um, search AI mode got an
01:05:13
upgrade to Gemini 3.5 queries that are
01:05:16
at an all-time high because users are
01:05:18
now treating search like a
01:05:19
conversational partner apparently.
01:05:21
>> Uh, allegedly.
01:05:22
>> Allegedly.
01:05:24
Uh, Sundar did say that like now people
01:05:26
are putting more natural language into
01:05:28
Google search which I have actually
01:05:30
found myself doing
01:05:32
>> because it used to be that we we knew
01:05:34
the language of SEO.
01:05:35
>> Everybody under a certain age grew up
01:05:38
learning how to Google something. Yeah.
01:05:40
Whether you had a class in it or you
01:05:42
learned some internet literacy somewhere
01:05:43
or from someone you know. So you have to
01:05:46
know what words to type into Google to
01:05:47
find what you're looking for. And then
01:05:48
also like how to exclude certain phrases
01:05:50
or like put things in quotes to find
01:05:51
like things together.
01:05:53
That's like a that's like a skill that
01:05:55
we all learned
01:05:56
>> that under a certain age people have not
01:05:59
had to learn that because they either
01:06:01
just go straight to an AI agent or an AI
01:06:04
and just type it in in natural language
01:06:06
and it finds it for them. And so now,
01:06:08
yeah, there's this like blurred line of
01:06:10
people who grew up not learning how to
01:06:11
Google things, but then they just use
01:06:13
natural language in Google and it kind
01:06:14
of works. It kind of works in AI mode
01:06:16
anyway and it's fine. It didn't used to
01:06:18
work at all and now you can you can do
01:06:20
that and it'll automatically do AI mode
01:06:22
and you can just keep talking to it.
01:06:23
Yeah.
01:06:24
>> Anecdotally speaking, I've been using
01:06:26
that a lot more just because the regular
01:06:27
SEO speak that I grew up using doesn't
01:06:29
work anymore.
01:06:30
>> That's true.
01:06:31
>> So, I've just been like trying to get
01:06:33
something. So, I'm like, "Okay, let me
01:06:34
just ask it the way I would normally ask
01:06:35
something."
01:06:36
>> Yeah. What's crazy now, though, is that
01:06:38
Google Search now has dynamic layouts uh
01:06:41
which can code custom visuals for you on
01:06:43
the fly. And it can also build sharable
01:06:46
mini apps like a weekend planner
01:06:48
itinerary. So how this was shown off was
01:06:51
basically I think it was Josh Woodward
01:06:53
that did this, but he was like, "Oh, I
01:06:54
want to do like this weekend thing with
01:06:56
my kids and my wife, but I'm not really
01:06:58
sure what to do. Like can you suggest me
01:07:01
things to do?" And so it builds this
01:07:03
little mini app inside of the Google
01:07:05
search window that shows the weather. It
01:07:07
shows like suggested locations. It shows
01:07:09
like dinner places they can go to and
01:07:12
stuff like that. and you can talk to
01:07:14
Google search to sort of modify it and
01:07:16
it will it will change the code of the
01:07:18
app in real time and you can share that
01:07:21
with whoever.
01:07:22
>> Um, which a part of this for me it was
01:07:25
like man people really want to offload
01:07:27
every decision they've ever had.
01:07:30
>> I I just like the idea of
01:07:32
then your like wife comes over like oh
01:07:34
did you figure out what we're going to
01:07:35
do this weekend? No, not yet. But check
01:07:37
out this app that's trying to figure out
01:07:38
what we're going to do. It's like if you
01:07:40
have a dog and like your dog is like
01:07:42
can't stop throwing up and so you take
01:07:44
your dog to the vet and then you come to
01:07:46
pick it up later and you're like, "So,
01:07:47
is my dog going to keep throwing up? Did
01:07:49
you fix that?" The vet's like, "I didn't
01:07:50
fix that." But your dog is literally a
01:07:51
transformer now. It's like this I still
01:07:54
have a puking shiny fur is.
01:07:55
>> That's how I feel about Google search.
01:07:57
>> Yeah. Yeah. It's strange.
01:07:59
>> I've Google searched so many things
01:08:00
during this show and the AI review has
01:08:02
been wrong about every single one of
01:08:03
them. And I'm just like I'm so sick of
01:08:05
this. Please do not build me apps.
01:08:08
Please just serve me relevant
01:08:10
information.
01:08:11
>> Yeah, I mean it's funny that we like
01:08:14
remember when they added in the
01:08:15
suggested snippets where it would like
01:08:18
give you the answer within Google search
01:08:20
and people freaked out because they're
01:08:21
like people aren't going to links any if
01:08:22
you do this. People won't go to the
01:08:23
links. And now they're like if a kid
01:08:25
searches how does a black hole work,
01:08:27
it'll just generate an animation on the
01:08:29
fly of a black hole colliding with
01:08:31
another black hole.
01:08:33
>> Yeah. which is on one hand on one hand
01:08:36
it's way worse the all the sources don't
01:08:38
get any traffic all of the people who
01:08:40
worked really hard on all the
01:08:41
information and research that made that
01:08:42
possible don't get any credit
01:08:44
>> but on the other hand
01:08:46
>> yeah it's cool
01:08:47
>> that kid just got an instant like
01:08:48
>> animation
01:08:49
>> explainer
01:08:50
>> if it's if it's correct
01:08:52
>> if it's correct
01:08:53
>> that also confused me though because
01:08:54
they said they would build it every time
01:08:56
on the fly yeah
01:08:58
>> there has to be some sort of like that's
01:08:59
a pretty normal question
01:09:01
>> well it's based on your specific speific
01:09:03
query.
01:09:03
>> But like let's say your query is how
01:09:05
does a black hole work?
01:09:06
>> Yeah, there's more than one pe one
01:09:08
person searching that every day. Like
01:09:10
why are you rebuilding that every time?
01:09:12
That has to be wasteful. Like why not
01:09:13
just
01:09:14
>> I think that's probably true, but
01:09:16
there's there's such a long tale of
01:09:18
unique Google searches every single day.
01:09:20
I think what they're thinking is in in
01:09:21
such a specific query which is like how
01:09:24
what would it look like if a black hole
01:09:25
the size of a a peanut collided with
01:09:28
Manhattan? And it's like boom, here's a
01:09:30
simulation of that. Which I don't know,
01:09:33
maybe three people Google that every
01:09:34
day. I don't know like how many people
01:09:36
say that, but there's like a ton of
01:09:37
unique versions of that question that it
01:09:39
can build a tailored animation just to
01:09:41
help you with cuz they have TPUs.
01:09:42
>> I've been trying to Google what is the
01:09:44
largest US data center by power
01:09:48
consumption, like by megawws. And it
01:09:50
gave me a list, right, which seems like
01:09:51
it's fine. The top five answers are
01:09:53
hallucinations on the list.
01:09:56
>> So, I was just like, what's the point of
01:09:57
this? I think these are more in like
01:09:59
longer wellestablished like if I this is
01:10:01
something I found too like when you ask
01:10:02
questions to Google about like current
01:10:05
events or recent information or recently
01:10:07
updated things it gets it wrong all of
01:10:08
the time. Yeah.
01:10:09
>> But if you ask it something about like a
01:10:11
10-year-old science theory or like like
01:10:13
an animal that everyone's been writing
01:10:15
books about or something that's not like
01:10:16
current events are brand new then you'll
01:10:19
get stuff that's pretty locks like
01:10:21
locked in like rock solid
01:10:23
>> right?
01:10:24
>> So it's it's the black hole stuff I
01:10:26
guess.
01:10:26
>> Yeah it is. It is strange. It feels like
01:10:28
Google search is they're kind of just
01:10:29
trying to nudge people towards the
01:10:32
process of using a Gemini search chat,
01:10:34
like a Gemini chat as opposed to
01:10:36
searching things anymore. So,
01:10:39
>> um, okay. Google AI Studio. Now you can
01:10:42
build full Android apps from scratch
01:10:44
with a prompt and you can publish them
01:10:46
directly to the Play Store, which means
01:10:47
we're going to get a lot of slop apps
01:10:49
that are terrible.
01:10:50
>> Apps, here they come.
01:10:51
>> So many. Next IO, they're going to brag
01:10:53
about how many new apps were created.
01:10:55
>> Guarantee it. This is so bad. Like they
01:10:59
should not let you publish these to the
01:11:00
Play Store.
01:11:01
>> Well, yeah. Will they meet a bar of
01:11:02
being approved? Cuz Google has a
01:11:04
famously lower bar of being approved.
01:11:06
Apple, it'll be like, "Oh, yeah, you can
01:11:07
use code to make apps." But it's not
01:11:08
like I hit a button and it generates an
01:11:10
app from scratch and then I submit it
01:11:11
and it's in the store already. Like
01:11:12
there's at least some threshold or
01:11:14
quality.
01:11:14
>> Yeah. But does that mean they're going
01:11:16
to have to hire like 50 times as many
01:11:18
approvers?
01:11:19
>> Yeah. Yeah. Yes, it does. Yeah. They're
01:11:21
just going to have one. Oh, true.
01:11:24
That's true. Hey, Gemini. Is this good
01:11:26
enough? You made it.
01:11:28
>> Also, I tried this yesterday and it did
01:11:29
not work.
01:11:30
>> Really?
01:11:31
>> Yeah. I mean, the app that's on your
01:11:33
Android phone is not out yet,
01:11:34
>> but they have like a web version of the
01:11:36
AI Studio thing and it just like wasn't
01:11:39
wasn't doing the thing. Cool.
01:11:40
>> What was the app you tried to make?
01:11:42
>> Um, I forgot. Wait, let me look it up.
01:11:44
The thing
01:11:45
>> people really have this many app ideas
01:11:46
all the time.
01:11:47
>> Well, okay. Well, mine was a dumb idea.
01:11:50
>> Well, I Yeah, I don't can't even think
01:11:52
of a cool dumb idea right now. I think
01:11:55
that what a lot of people might do in
01:11:57
the future is like there are a lot of
01:11:59
apps that are paywalled and if you're
01:12:01
just like I want a calorie tracker app.
01:12:03
>> So stealing.
01:12:05
>> Yeah. That's what all this is. It's all
01:12:07
stealing.
01:12:07
>> It's all stealing.
01:12:09
>> It's also people want
01:12:11
>> a lot of people want apps that are
01:12:13
specifically tailored to them. So like
01:12:15
there's a bunch of generic versions of
01:12:17
apps that do something. Uh, I won't say
01:12:19
his name, but I have a teammate who's
01:12:20
like, I I want an app that can like tell
01:12:23
me specifically if I need to wear
01:12:26
sunscreen today. And all it does is
01:12:28
intake like how long I'm going to be
01:12:29
outside, what the weather is, what my
01:12:31
skin tone is, and all these other
01:12:33
things. And so just like there's no app
01:12:34
that does that, so just build me the app
01:12:36
and then it'll just do it for you. Where
01:12:38
you could probably find an app that does
01:12:39
like twothirds of what you're asking,
01:12:40
but
01:12:41
>> then is the developer going to do the
01:12:43
extra feature you want? I don't know. So
01:12:45
just it's or maybe you want a task
01:12:47
manager app and there's a certain
01:12:49
feature. Oh, I just want it to be like
01:12:50
this, but also with this extra reminders
01:12:51
thing where it's geotagged and it
01:12:53
reminds me to do something when I get to
01:12:54
a certain place. Whatever. I want an app
01:12:55
just for me.
01:12:57
>> Then this is your way of making an app
01:12:58
just for you. So like why put it on the
01:13:00
Play Store? I don't know.
01:13:02
>> But that's why I see people making apps.
01:13:04
>> Yeah,
01:13:05
>> Adam made us local frame.io. That's just
01:13:07
like
01:13:08
>> video player big clock add comments
01:13:11
export comments as CSV.
01:13:13
>> It's perfect. something that just for
01:13:15
us, we would use it. We would never put
01:13:18
it on the Play Store or whatever, but
01:13:20
like it's useful for us.
01:13:20
>> I mean, one thing that's cool about
01:13:21
this, too, is that you can export the
01:13:24
app and send it to just friends and
01:13:25
family without putting on the Play
01:13:27
Store.
01:13:28
>> So, I think that's a sort of interesting
01:13:29
idea of making like family apps or
01:13:31
friends apps and stuff like that.
01:13:33
>> Oo, a group chat. Oh, there's no
01:13:35
bubbles. Never mind.
01:13:36
>> Wait, can we do group chat?
01:13:37
>> Make them blue.
01:13:38
>> Turn the bubbles.
01:13:40
>> I found my um my app idea from
01:13:42
yesterday. Oh,
01:13:43
>> and it also, this is a late episode. Did
01:13:46
they even test this? Because this goes
01:13:48
back to an issue I've been having. So on
01:13:50
Fitbit, they have meditation minutes.
01:13:52
And if you try to meditate for 10
01:13:54
minutes with a Pixel Watch, it'll
01:13:56
vibrate on your wrist for the entire 10
01:13:58
minutes so that you know to breathe
01:14:00
along with it. It's like breathe in.
01:14:02
>> Oh my god.
01:14:03
>> Breathe out.
01:14:05
>> Like they got way too cute with it. I
01:14:07
just want a timer that will vibrate at
01:14:08
the end of 10 minutes to let me know
01:14:10
you're done.
01:14:10
>> Probably tanks the battery life, too.
01:14:12
probably, but they don't have that. So,
01:14:14
I tried to have uh AI Studio build it
01:14:16
for me, and instead it built me an
01:14:17
interactive dashboard where I can play
01:14:19
with a fake Pixel Watch 3,
01:14:23
>> which is technically really impressive,
01:14:25
but not useful at all.
01:14:27
>> Build a timer. You said build a
01:14:29
10-minute timer and it said
01:14:31
>> Yeah.
01:14:32
>> What?
01:14:32
>> I think because I said for my Pixel
01:14:34
Watch 3, it freaked out and it it built
01:14:36
a whole other thing. So,
01:14:37
>> you might be on to something though with
01:14:39
the group fitness challenges. Can we
01:14:41
finally have our own group fitness
01:14:43
challenge?
01:14:43
>> This is something we've wanted. Well, I
01:14:45
think we we should this is the perfect
01:14:47
ideal for us. We should be able to build
01:14:49
a multi-platform group fitness
01:14:51
challenges app
01:14:53
>> that in our studio we can distribute it
01:14:55
among us and we can use whatever
01:14:56
wearable we want.
01:14:57
>> Yeah.
01:14:58
>> And it'll normalize all of our stats,
01:15:01
>> calories.
01:15:02
>> I built it. I can make it so I always be
01:15:04
market.
01:15:04
>> No, that's
01:15:06
I will look in your code. I will inspect
01:15:08
your code with my agent. to find out if
01:15:10
you're cheating. Agent, look at his
01:15:12
code.
01:15:13
>> Yeah,
01:15:14
>> peer review.
01:15:14
>> Anti-gravity. Look at his anti-gravity.
01:15:17
>> Peer reviewed.
01:15:18
>> Uh, okay. Google pics.
01:15:22
>> We referenced this at the beginning.
01:15:23
>> What a transition.
01:15:24
>> Can you stop Google? Can you Can you
01:15:26
stop Google photos, Google Pics, Google
01:15:28
images?
01:15:29
>> They literally they they showed the
01:15:30
slide. They're like, "This is Google
01:15:32
Pics." And I was like, "Are you sure you
01:15:34
want to manage that?
01:15:35
>> Sure.
01:15:35
>> Why are you doing this?" Uh, this is a
01:15:37
new text and hover image editing tool.
01:15:40
Uh, basically it's sort of similar to we
01:15:43
talked about Google books a little bit
01:15:44
last week. How you can sort of like
01:15:46
hover the mouse cursor over things and
01:15:47
it sort of intelligently knows what you
01:15:49
want to do with it.
01:15:50
>> That's not Google books. That's the
01:15:51
Google book.
01:15:52
>> That's the right cuz Google books.
01:15:54
>> Not to be confused with Google Books.
01:15:55
>> Yeah. Google Books.
01:15:56
>> Kindle. Yeah. It's the Kindle thing.
01:15:58
>> Yeah.
01:15:59
>> No, that's that's books on the Google
01:16:01
Play Store. Google Books is the online
01:16:03
literature database. No, that's you're
01:16:05
thinking of Google Play Books.
01:16:06
>> Yeah. No, no. Google Play Books is the
01:16:08
book is the Kindle.
01:16:09
>> Isn't the online database literature
01:16:11
thing like
01:16:12
>> that's called Google Books?
01:16:13
>> Is it?
01:16:13
>> Yeah.
01:16:14
>> Okay.
01:16:14
>> Then there's also Google Scholar.
01:16:16
>> Scholar. That's what I was saying.
01:16:17
>> Google Scholar is also different.
01:16:19
>> That's another thing. That's all the
01:16:20
papers and and research and PDFs. This
01:16:23
is this is untenable.
01:16:25
>> Do you know that the one it's called
01:16:26
Google get?
01:16:30
>> All right. Anyway, uh so it's this text
01:16:33
and hover image editing tool. Basically,
01:16:36
you take an image in and you can say
01:16:37
like zoom out and it'll like add extra.
01:16:40
>> It's a photo editor.
01:16:41
>> It's a photo editor.
01:16:42
>> Why is it separate from
01:16:44
>> Why is it separate from Gemini? This is
01:16:46
my question about every single one of
01:16:47
these products.
01:16:49
>> Imagen is a Google DeepMind product.
01:16:51
They're on 4 now.
01:16:52
>> Yeah. Sorry, I was making another punt
01:16:55
because you said you can take an image
01:16:56
in and I was like is which is true. Imag
01:16:59
is actually a separate Google product.
01:17:01
>> Yes, correct. I
01:17:02
>> But yeah, but
01:17:05
>> so why do they have 50 of these?
01:17:08
>> Why is this not in Google Photos?
01:17:09
>> Why don't they just say that's the first
01:17:11
I as we were watching it I was like so
01:17:12
is this going to be built into Google
01:17:13
Photos or is this separate? Google Pix
01:17:16
is separate from Google Photos which is
01:17:18
separate from Google
01:17:18
>> or literally they can just say Gemini
01:17:20
can now do this other thing now. Like I
01:17:22
don't understand.
01:17:24
>> Why are they spinning them off into a
01:17:26
ton of different products? cuz this team
01:17:27
is in a different building.
01:17:29
>> Oh, I mean last year at IO they they
01:17:31
announced this big product and they
01:17:32
sunset it last month. Like I I bet you
01:17:34
within a year 80% of these are going to
01:17:36
be shut down.
01:17:38
>> The people who make the Google logos are
01:17:40
just like
01:17:41
>> check finally finished my backlog and
01:17:44
then they look at the just 25 new names
01:17:46
that pops out of Google IO and they're
01:17:48
like oh my god.
01:17:48
>> They go to thesaurus and they look at
01:17:50
photo and they're like what other
01:17:51
products do you use for every single
01:17:54
thesaurus? Okay. Um, okay. They have
01:17:57
Google Stitch and Google Flow now. They
01:17:59
already had these, I guess, but they're
01:18:01
updated. Stitch was a website builder
01:18:03
and now it's updated and um, yeah, it
01:18:07
can build you websites in real time. And
01:18:08
then Google Flow uh, allows creators to
01:18:11
edit 16 video seen simultaneously. But
01:18:14
there's also Google Flow music
01:18:18
which can take like they gave it an
01:18:20
example of they played a piano riff and
01:18:22
the guy was like what would this piano
01:18:23
riff sound like if there was a bass
01:18:25
guitar and a singer?
01:18:26
>> This is insane. It was just another
01:18:28
insane demo that somehow has a name and
01:18:30
a product.
01:18:31
>> Yeah. Like
01:18:33
>> which you should just be able to put the
01:18:34
audio clip in Gemini and say do this.
01:18:37
>> Yeah. We're getting to the point
01:18:38
>> in IO where they kept cutting to the
01:18:40
audience and I've never seen a group of
01:18:43
people who were more bored. Like the
01:18:45
amount of just looking down at phones
01:18:47
that happened by at minute 90, it was
01:18:51
rough.
01:18:51
>> It was really funny. Yeah, I I meant to
01:18:53
bring that up. They kept cutting to the
01:18:54
audience like every 30 seconds and and
01:18:56
the first 20 minutes of the show, they
01:18:58
were cheering for stuff and they were
01:19:00
pretty pumped and Sundar would be like
01:19:01
and a trillion tokens and they'd be
01:19:03
like, "Wow." By minute 70,
01:19:07
they would have like pause for applause
01:19:10
on the teleprompter. You could tell and
01:19:11
it would just be a few seconds of
01:19:13
silence and then
01:19:17
>> and then they would just move on and
01:19:18
you'd cut to the crowd and they'd just
01:19:19
be like on their phones trying to figure
01:19:21
out what's going on. There was one point
01:19:22
where the presenter said, "And you can
01:19:23
clap for this."
01:19:24
>> Yeah.
01:19:25
>> And I was like, "Oh, that's painful."
01:19:28
>> All right. Uh, basically last thing. Oh,
01:19:30
no, cuz we'll get to the glasses. But
01:19:31
last thing here with the app and search
01:19:33
overhauls, synth ID and C2PA. Uh if you
01:19:36
don't know what C2PA is, it's content
01:19:38
credentials. We made a dedicated video
01:19:40
on the studio channel about it that is
01:19:41
also tied in with an M11P video. You can
01:19:44
go watch that if you want to know more
01:19:45
about it. Uh but synth ID is Google's AI
01:19:49
watermarking feature where basically
01:19:51
every piece of AI generated content that
01:19:54
they make has like an invisible
01:19:56
watermark in the metadata that other AI
01:19:58
can well that previously only Gemini
01:20:00
could read and it would tell you whether
01:20:02
or not something was AI. They are
01:20:04
expanding that significantly to
01:20:06
basically every company which is really
01:20:09
good. Like that that's something that's
01:20:10
very important. It's very ironic that
01:20:13
they're creating the exact thing that
01:20:15
you now have to uh check. There was a
01:20:17
there was a part where I think it was
01:20:19
Sundar was like only one in four people
01:20:22
can tell when a piece of content is AI
01:20:24
generated.
01:20:26
>> And that's why we made s like made synth
01:20:29
ID. It has like the same shades of like
01:20:32
>> we've made this this smartphone
01:20:34
extremely like
01:20:36
>> addictive
01:20:37
>> and so we have this digital wellbeing
01:20:39
feature that you can use to use your
01:20:40
phone less. So like, oh yeah, we we've
01:20:42
made this AI generated content extremely
01:20:45
realistic.
01:20:46
>> And so now we've developed another AI
01:20:48
tool to help you decide if it's real or
01:20:49
not.
01:20:50
>> I think a perfect example of what we
01:20:51
talked about before is like the problem
01:20:53
with this is we created the tool that
01:20:55
can spread a ton of misinformation, but
01:20:56
also the tool that can prove it
01:20:58
misinformation. The problem is is the 10
01:21:00
million people who saw the
01:21:01
misinformation,
01:21:03
>> 100,000 of them might see the watermark.
01:21:06
The Sony tweet that you said before that
01:21:08
had 13 million views. The follow-up
01:21:10
tweet that explained it a little more
01:21:12
had 300,000 views. Like people aren't
01:21:14
looking for the update. They're reading
01:21:17
the title, watching the image, seeing
01:21:18
the video, and now that is fact.
01:21:21
>> And to be clear, it's also always in the
01:21:23
context. So this is a it's not a visible
01:21:25
watermark. It's basically buried in an
01:21:27
invisible form so that when you scroll
01:21:29
past this on your feed, you aren't
01:21:31
thinking, "Oh, let me go drag this and
01:21:34
check this to see if it's AI or not."
01:21:37
>> When you just see it in your feed
01:21:38
randomly, you're not in the in the
01:21:40
context of like thinking, is this AI or
01:21:42
not? So, it just doesn't get checked.
01:21:45
Which is why, and this is the first time
01:21:46
I'm ever saying this and the last time I
01:21:48
will ever say it, Meta actually had a
01:21:50
good idea. And on Instagram, they put
01:21:52
the AI info thing. Although AI info is
01:21:55
terrible verbiage and they should just
01:21:56
have edited with AI or modified with AI
01:21:58
or generated with AI. Uh but that wasing
01:22:01
up because if you used Photoshop at all
01:22:03
it would say AI info which is all broken
01:22:06
but the platforms themselves need to
01:22:08
support synth ID. Another rare statement
01:22:11
to Twitter's credit.
01:22:13
>> Now this is where it gets weird. They
01:22:15
tried to uh implement this thing where
01:22:17
it autodetects if the image you're
01:22:19
attaching is made with AI or not. And if
01:22:21
it is, then it has it adds a little tag
01:22:22
so that in the feed as you're scrolling,
01:22:24
you don't have to think about it. It'll
01:22:25
just say made with AI.
01:22:26
>> Yeah.
01:22:27
>> Unfortunately, there's a ton of false
01:22:28
positives. And I uploaded images that
01:22:30
are not made with AI that it just tags
01:22:32
them as made with AI. So, I'm not sure
01:22:34
how it's reading that, if it's just
01:22:35
trying to look at the image and decide
01:22:37
or if it's using C2PA or some
01:22:39
combination of that. I don't know. But
01:22:40
there's a ton of false positives.
01:22:41
>> And it it needs to go a step further to
01:22:44
bandwagon off. It can't just be like
01:22:46
accurate, reliable synth ID usage. It
01:22:48
needs to be like an o it can't be like a
01:22:50
black box because when you have these
01:22:52
platforms that are acting as publishers
01:22:54
if hypothetically there's a politician
01:22:57
that the owner of Twitter does not like
01:23:01
there's nothing stopping them from just
01:23:02
slapping the AI generated label on all
01:23:04
of their
01:23:04
>> or the other way around on their
01:23:05
something is AI generated from they do
01:23:08
like and taking it away.
01:23:09
>> Yeah. The only reason it has to be a
01:23:10
black box is because of bad actors,
01:23:13
>> like outside bad actors who want to
01:23:15
abuse and show things on the feed and
01:23:18
get through the algorithm without the
01:23:19
tag.
01:23:19
>> Well, that's why C2PA has like a it has
01:23:22
a trail that you can actually follow.
01:23:23
Like when something has content
01:23:24
credentials on it and you pull it into
01:23:26
like the content credentials checker
01:23:28
thing, it shows you every edit that's
01:23:29
been added and when that edit happened.
01:23:32
>> But I don't think that synth ID has
01:23:34
that. It just shows it just tells you if
01:23:35
it's been if it's been made with AI or
01:23:38
not. But either way, it is really good
01:23:40
that a ton of services are picking up on
01:23:42
synth ID. Um, it's just very ironic that
01:23:44
it's basically like Google's like, "Look
01:23:46
at all this slop that you can make and
01:23:48
trick people." And then there was also
01:23:50
they they had a bunch of influencers
01:23:52
like bake digital twins of themselves
01:23:53
with Gemini Omni.
01:23:55
>> I did not like this.
01:23:56
>> It was so weird. like they had a a
01:24:00
Disney adult uh basically talking about
01:24:03
her favorite Disneyland land foods and
01:24:05
then they had her climbing a rock and it
01:24:08
was just very uh it was very like it was
01:24:11
just creepy and weird and um we
01:24:14
shouldn't do that slop generators.
01:24:17
>> Before we move on uh away from AI
01:24:19
products and towards wearables,
01:24:20
>> I can't wait till we move. Um, I just
01:24:22
want to say I think I think we're going
01:24:23
to see a lot more of this over the next
01:24:25
year because
01:24:27
>> there's so much like circular money in
01:24:30
the AI space and there's so much money
01:24:32
committed to these these data centers. I
01:24:34
did finally
01:24:35
>> find a a list and uh and the most
01:24:38
important number is that like right now
01:24:39
in the US we have about 18 18 gawatt
01:24:43
operational. In the next like 10 years
01:24:45
we're planning on building 334 gawatt.
01:24:49
Um, for perspective, one gigawatt is
01:24:52
about as much as a large nuclear power
01:24:54
plant produces. Um, and so like all of
01:24:57
the companies like burning through these
01:24:59
tokens for a large part are doing this
01:25:01
on like subsidized money because
01:25:03
companies like Nvidia are funding these
01:25:07
like startups that are then buying
01:25:08
tokens with the Nvidia money. So all the
01:25:10
companies like Google that have are
01:25:14
starting to
01:25:16
in what what's what do they say like
01:25:17
their declarations of intent to like
01:25:19
spend all this money on data centers
01:25:21
they need to find customers that they
01:25:23
are not paying and like fast because
01:25:26
otherwise like the dates these starting
01:25:28
construction dates are going to start
01:25:29
coming around and they're not going to
01:25:31
have the cash flow to again
01:25:33
>> build hundreds of times of the electric
01:25:36
you know like
01:25:37
>> I mean open AI in particular right? Like
01:25:40
they they have basically said like, "Oh
01:25:42
yeah, like our numbers don't make sense,
01:25:44
but next quarter we're going to add 30
01:25:46
350 billion new users."
01:25:48
>> Yeah. So we're we're going to see just
01:25:50
like tons and tons and tons of like
01:25:52
these these like AI product cases in
01:25:55
just like a desperate attempt of like,
01:25:57
"Hey, can you please please buy some
01:26:00
tokens that we didn't pay you to buy?"
01:26:02
>> Right.
01:26:02
>> Have you guys ever heard of the term
01:26:04
Ponzi scheme?
01:26:06
>> It seems really familiar right now.
01:26:08
Yeah, it's just that the problem right
01:26:10
now is they're building so much
01:26:11
infrastructure, but they just are not
01:26:13
seeing the demand. And they're having a
01:26:15
lot of demand and they're saying, "Oh,
01:26:17
like look at all this demand. We need to
01:26:19
build the infrastructure." But they
01:26:20
they're building so much infrastructure
01:26:22
because they say the demand is going to
01:26:24
scale to that within the next 10 years,
01:26:26
but the numbers are not showing that it
01:26:28
actually is going to. And that's why
01:26:29
Gemini is getting put in every single
01:26:31
product because it creates demand that
01:26:33
people didn't ask for.
01:26:35
>> And usage.
01:26:36
>> That's what I mean. Yeah. Like
01:26:37
artificial demand. getting rid of the
01:26:38
thing you're actually using and
01:26:40
replacing it with the AI overview and
01:26:42
then being like, "Look at all the people
01:26:43
that are using AI overview."
01:26:44
>> Yeah, they actually clearly did that on
01:26:46
stage where they were like, "Yeah, we've
01:26:47
never had as many people using AI mode
01:26:49
in Google search." And I was like,
01:26:51
>> mostly been using it by accident.
01:26:53
>> Yeah. Well, that's it like automatically
01:26:55
AI modes for you.
01:26:56
>> Yeah.
01:26:57
>> So, obviously, we have a lot of usage
01:27:00
>> this year. This year, we've seen a
01:27:02
record number of people fall down the
01:27:04
stairs. Coincidentally, we removed all
01:27:06
handrails.
01:27:06
>> Yeah. It's Alison Johnson wrote a really
01:27:08
good piece at the Verge right before
01:27:10
Google IO about how Gemini is getting
01:27:12
way too close to being like in your face
01:27:14
a little too much. Like at first it was
01:27:17
like, "Oh yeah, I can use Gemini for
01:27:19
that." But now that they've put Gemini
01:27:20
in every single Google product,
01:27:22
>> it's a search box.
01:27:23
>> The logo is everywhere. It's like, do
01:27:25
you want to do Gemini? You want to do
01:27:26
Gemini? You want to do that with Gemini?
01:27:27
>> Dude, opening a Google doc, it just like
01:27:29
surrounds your doc.
01:27:30
>> There's something at the bottom of our
01:27:31
page right now I've never seen before.
01:27:33
>> It's a new little toolar any changes you
01:27:35
want to make. It tries to like edit your
01:27:36
document for you. There's the Gemini
01:27:38
logo in the top right
01:27:40
>> where you can do the same thing.
01:27:41
>> Dude, if you open Google Docs, it's just
01:27:43
like
01:27:44
>> they are really trying to use like as
01:27:46
much AI as possible. It's in the top
01:27:47
right corner. It's in the bottom corner
01:27:49
>> because they have to gloat about how
01:27:50
much people are using Gemini at the next
01:27:51
Google IO. This is their whole thing.
01:27:53
>> Yeah.
01:27:54
>> Anyway, okay, we're going to talk about
01:27:56
the smart glasses now.
01:27:57
>> Yeah. So, we can talk about the glasses.
01:27:58
The last bit really with the glasses is
01:28:00
they saved it for near the end but they
01:28:03
unveiled uh that they would have what
01:28:05
they called audio glasses but it's
01:28:08
essentially the same product as Meta's
01:28:10
Ray-B bands but with Google's AI and
01:28:13
Gemini instead. Yeah. So with Warby
01:28:15
Parker, I think, and Gentle Monster,
01:28:17
they have like these two form factors of
01:28:18
these glasses with cameras on the front
01:28:20
with speakers, with microphones, with
01:28:22
the compute on them, and you can talk to
01:28:24
Gemini, which of course is linked with
01:28:26
your phone and can do a bunch of things
01:28:27
like deploying agents and all that fun
01:28:29
stuff that they talked about through the
01:28:30
whole keynote. You can do all of that
01:28:32
with just wearing the glasses.
01:28:33
>> Yeah.
01:28:34
>> So yeah, I don't know why they're called
01:28:36
audio glasses. Maybe because
01:28:37
>> cuz now they're going to have audio.
01:28:39
They're going to have the screen
01:28:40
glasses. So audio is the main way that
01:28:42
you're interacting with Gemini since
01:28:43
there is no display. They will for sure
01:28:45
at some point have a moninocular and
01:28:47
binocular version of like a display
01:28:49
glasses. Maybe that's what they're
01:28:50
called
01:28:51
>> next year.
01:28:51
>> So that that that'll be coming.
01:28:53
>> Visual glasses.
01:28:54
>> Yeah. But uh yeah, these are these are
01:28:56
the audio glasses and they are I I guess
01:28:59
in collab with Samsung and and wear OS
01:29:02
where
01:29:03
>> Android XR
01:29:04
>> and and Android XR.
01:29:06
>> Yeah. Gemini.
01:29:07
>> Yeah. They had some cool demos actually
01:29:09
though. the they had the lady come out
01:29:10
and she was she had the glasses on and
01:29:12
she was like
01:29:13
>> hey can you direct me to that place that
01:29:14
I went with Allison last week and it
01:29:16
knows exactly where that was and I think
01:29:18
that's kind of cool and then it's like
01:29:20
do you want to stop for your cold brew
01:29:22
on the way that you usually get which I
01:29:23
thought was kind of interesting and
01:29:25
again making more money for the
01:29:27
advertisers
01:29:28
>> everything is like how do I get your
01:29:30
credit card information
01:29:32
>> Google is a fishing scam at this point
01:29:35
like just everything they do just wants
01:29:37
me they want my credit card information.
01:29:39
>> Yeah, for sure. Yeah. And then um as
01:29:41
she's like walking there, it's like
01:29:43
she's like, "Oh, hey, can you can you
01:29:44
order it for me so it's ready when I get
01:29:46
there?" And then it you agentically uses
01:29:48
her phone and like taps through the
01:29:50
things and uses Door Dash to like order
01:29:52
it for her, which I was like, I guess
01:29:54
nobody's going to be talking to the
01:29:55
Breezas anymore. It's like what happened
01:29:57
with Starbucks when Starbucks introduced
01:29:59
their app. Yeah. And it's like tons of
01:30:01
people will be waiting in line and then
01:30:03
people will just come in and get their
01:30:04
coffee because they had to prioritize
01:30:06
the mobile orders because it was a new
01:30:08
feature that they needed people to use
01:30:10
>> and it was probably most of their
01:30:11
business.
01:30:11
>> It was it probably still is most of
01:30:13
their business. Yeah,
01:30:14
>> I feel like we had it. We had that
01:30:15
happen to us. Remember? I don't remember
01:30:16
where we were at some some event where
01:30:19
like we were at the front of the line
01:30:20
for a solid five minutes and no barista
01:30:22
talked to us or even acknowledged us as
01:30:24
they just went back and forth making
01:30:26
mobile orders and people walked in,
01:30:28
picked up the food and left and we were
01:30:29
at the front of the line for like
01:30:30
several minutes
01:30:31
>> order in the store.
01:30:32
>> Yeah,
01:30:33
>> I'm so tired of everyone drive-thru.
01:30:36
>> Are you using the app today? I would
01:30:37
tell you if I was using the app today,
01:30:39
please just take my order.
01:30:40
>> I don't know why it has to ask me every
01:30:42
single time.
01:30:42
>> I forget that they have apps for for
01:30:44
fast food now. That's great.
01:30:45
>> Which is fine. I think that's like I
01:30:47
don't have a pro. I don't use them, but
01:30:48
I don't have a problem with them. But
01:30:50
you don't you can just say hello.
01:30:53
>> I use some apps,
01:30:54
>> but drive-through should first say
01:30:56
>> we're getting we're now
01:30:56
>> what can I get for you?
01:30:58
>> Or like how are you today? And then I
01:30:59
could say oh I have an order on the app.
01:31:02
>> Yeah.
01:31:03
>> And then they'd give you a fee. Yeah.
01:31:04
Something I don't think that Google is
01:31:06
quite prepared for is so you there's a
01:31:10
lot of negativity around the meta
01:31:11
glasses, you know, cuz people in general
01:31:13
hate AI, but people especially hate
01:31:16
being filmed without their consent.
01:31:18
>> That's the main thing.
01:31:19
>> And then stuff being put online about
01:31:20
it. And I've seen there's clips of
01:31:22
people who are like people like, "Hey,
01:31:24
are those metagasses? Are you recording
01:31:26
me right now? Are you?" And Google
01:31:28
basically went through this with Google
01:31:30
Glass 1.0 because they didn't put a
01:31:31
recording indicator on the front. Mhm.
01:31:33
>> I really, really, really hope that
01:31:35
they're dictating that every Android XR
01:31:37
glass device needs to have a recording
01:31:39
indicator.
01:31:40
>> They didn't talk much about it.
01:31:42
>> They didn't.
01:31:42
>> It's got to.
01:31:43
>> I hope so. But I just am a little bit
01:31:46
worried that the more because right now
01:31:48
there's basically the meta glasses and
01:31:49
that's basically it. There's like some
01:31:50
other smart glasses, but nobody uses
01:31:52
ones.
01:31:52
>> Yeah. But like nobody buys those. So,
01:31:55
I'm a little concerned that like once
01:31:57
because the Android ecosystem of smart
01:31:59
glasses, which will be a ton of
01:32:00
companies at some point,
01:32:02
>> once those come out, no, everyone's
01:32:04
going to be like, "Am I being recorded
01:32:06
right now?" all the time,
01:32:08
>> right? Because at least when someone
01:32:09
picks up their phone and records you,
01:32:11
it's obvious to you.
01:32:12
>> Yeah.
01:32:13
>> But the amount of unwanted recording is
01:32:15
going to go way up and the sentiment
01:32:16
around AI glasses is going to go way
01:32:19
down. I wonder if I mean I don't know if
01:32:21
they would do this but I wonder if it's
01:32:23
possible for Google to add some sort of
01:32:25
a require a hardware requirement for XR
01:32:29
glasses. So it's like if you want to use
01:32:31
Android XR you must have insert minimum
01:32:34
resolution here insert cameras here
01:32:37
insert like indicator recording light.
01:32:39
>> That would be neat.
01:32:41
>> Yeah, I hope so.
01:32:42
>> It should be also when the glasses are
01:32:45
recording when you didn't ask them to.
01:32:47
anytime. Yeah. Anytime that sensor is
01:32:50
detecting light, period, a light should
01:32:52
be,
01:32:52
>> which is something that
01:32:54
>> feels obvious.
01:32:54
>> Pixels do, right? Or is it Androids do
01:32:57
where like you can have the indicator
01:32:58
come on that any single time that the
01:33:00
microphone or the camera is turned on,
01:33:02
there is an indicator in your status
01:33:03
bar.
01:33:04
>> Yeah, that's an Android feature. So, I
01:33:05
hope Android XR has that as well.
01:33:06
>> Yeah. Yeah. Um,
01:33:09
weirdly enough, like when Meta released
01:33:10
the Ray-B bands and they were trying to
01:33:12
get you to use the AI features and like
01:33:13
literally nobody uses them, I always
01:33:16
thought that the AI stuff was stupid.
01:33:18
Uh, but that's mostly because I don't
01:33:19
trust Meta and I don't trust meta AI.
01:33:21
But because you use Google all the time,
01:33:24
I actually see a lot of use cases where
01:33:27
I have a thought about because I I take
01:33:28
photos about things and then search them
01:33:30
all the time. Mhm.
01:33:31
>> So, being able to do that and knowing
01:33:33
it's Google on the back end, I feel a
01:33:35
lot more confident in it.
01:33:37
>> So, I would actually probably use this
01:33:39
type of product a lot.
01:33:41
>> Yeah.
01:33:41
>> Uh the problem is the cameras.
01:33:44
>> Yeah. It depends on how how plugged in
01:33:46
you're willing to be.
01:33:47
>> Like imagine imagine the magic Q moment
01:33:49
where you don't even have a display, but
01:33:50
someone texts you and goes, "Hey,
01:33:52
remember that picture from uh the crane
01:33:54
on top of Marquez's Tesla?" And you're
01:33:56
like, and the the assistant on your
01:33:58
glasses goes, "Do you want to just send
01:34:00
him the picture?"
01:34:01
>> The doesn't even show you the picture.
01:34:03
Do you trust it that much?
01:34:05
>> Do you trust it that much?
01:34:06
>> And then and then it's like, Mto's
01:34:08
reply, "What the why would you send me
01:34:11
that?"
01:34:12
>> Well, imagine you ask your glasses to
01:34:14
buy you your cappuccino and you get
01:34:16
there and you've got like a triple
01:34:17
deluxe ham sandwich or something, you
01:34:20
know?
01:34:20
>> Yeah.
01:34:20
>> I mean, you got to take a picture of it
01:34:22
with your soaked experience. close it up
01:34:25
>> or like worse all of the stuff you just
01:34:28
said works on the first try and then
01:34:31
you're just this weird meat vessel for a
01:34:34
Google data center somewhere like which
01:34:36
is exact like picture like then like a
01:34:38
few days later being like yo thanks for
01:34:39
sending me that pic and you're like I
01:34:41
don't know what you're talking about
01:34:43
>> I just wear the glasses from
01:34:48
like you just arrive at the coffee shop
01:34:51
you pick up your like why am I here
01:34:53
>> cuz it suggest you things non-stop and
01:34:55
you just say yes or no. It's
01:34:56
>> like turn left. You're like, why am I
01:34:57
turning left? Where am I going?
01:34:59
>> Coffee shop.
01:35:00
>> I feel like a lot of these tech
01:35:01
companies used to try to solve problems
01:35:02
and now they're just
01:35:04
>> both.
01:35:04
>> No. I feel like they're trying to like
01:35:06
guide user behavior instead of answer
01:35:08
>> our issues
01:35:09
>> 100%.
01:35:10
>> Any of you guys seen Star Trek 2: Wrath
01:35:12
of Con? Yeah. It's like the brainworm
01:35:14
thing. Like
01:35:15
>> it is there is this weird thing where
01:35:16
like I think a lot of tech companies
01:35:19
believe that people don't know what they
01:35:21
want until you show it to them which was
01:35:26
true for a long time in a lot of things.
01:35:28
But now with these convenience features
01:35:30
it's almost like you're guiding people
01:35:32
along with things that come on you
01:35:35
definitely want this. You definitely
01:35:36
want us to to order that thing for you
01:35:38
and like direct you to the thing that
01:35:40
you usually do. And it just becomes like
01:35:42
you sort of just following along the AI
01:35:44
instead of making your own decisions.
01:35:45
>> And they want you to do that because
01:35:46
then they can go to advertisers and say,
01:35:48
"Hey, we can sell your product to
01:35:50
people." Yeah.
01:35:50
>> Look how good we are. You don't you
01:35:51
shouldn't use meta. You should use
01:35:53
Google shopping.
01:35:53
>> And then it'll all have been worth it.
01:35:55
>> Yeah. And then all the AI investment
01:35:57
will be worth it.
01:35:58
>> Um Okay.
01:35:58
>> Are we old?
01:36:00
>> Yeah.
01:36:01
>> Okay. You want to Okay.
01:36:02
>> This whole time I've been like, I hit
01:36:05
there's the line, we hit it. I'm the
01:36:07
boomer now. like kids are going to be
01:36:09
like you just get with the time gramps
01:36:12
and like this is the line that I'm not
01:36:13
with the times.
01:36:14
>> It's weird because a lot of like in my
01:36:16
life I feel like I'm perfectly in the
01:36:18
middle of regular people and then very
01:36:21
much technology people and everyone on
01:36:23
the technology side seems to be very
01:36:25
excited for the AI and then everyone in
01:36:27
my regular life is like this is the
01:36:29
worst thing ever. It's going to kill us
01:36:30
all. So we we kind of like tow that line
01:36:33
and I feel myself going I'm excited for
01:36:35
it but I'm scared of it but it's cool
01:36:37
but it's like why are we doing this
01:36:39
>> flip-flopping all over optimistic about
01:36:41
AI while totally sick of what's going on
01:36:44
with it right now and exhausted by it.
01:36:47
>> I want to make videos about like you
01:36:49
know AI stuff on my channel but not like
01:36:53
AI AI. I want to make videos about like
01:36:54
>> top five prompts too.
01:36:56
>> No not like that. I want to make like
01:36:58
theoretical thoughts I deployed this
01:37:01
week.
01:37:02
>> Yeah. More more like what happens if
01:37:05
this AI thing works and like what do
01:37:07
what does that say about society?
01:37:09
Whatever. But
01:37:10
>> AI does kind of all get lumped together.
01:37:12
>> That's the problem.
01:37:13
>> One big AI.
01:37:14
>> Yeah.
01:37:14
>> So if there are little pieces of it that
01:37:16
are that are great like the tools we
01:37:18
talk about all the time that we may use
01:37:19
whether it's like oh we got to make like
01:37:21
a little frame. IO that's for us that
01:37:22
has special features or even just like
01:37:24
help me cut something out of Photoshop
01:37:25
easier. that gets something to the same
01:37:27
AI as like all the chat bots and all the
01:37:28
other things that are happening and and
01:37:29
all the AI on the glasses and it's all
01:37:31
AI so it's all
01:37:33
>> has this label so it just kind of gets
01:37:35
locked. It's like the same AI that lets
01:37:37
you uh generate an a useful widget on
01:37:39
Android to help you track your run and
01:37:41
how many things you're going to do
01:37:42
before the marathon
01:37:44
>> gets lumped into all the rest of the
01:37:46
things you've heard about AI for the
01:37:47
past hour.
01:37:48
>> Right. Yeah, that's the problem. We've
01:37:50
been talking for way more than an hour.
01:37:52
>> Yeah. I think a lot of people just feel
01:37:53
really insulted given how many like
01:37:56
serious flaws there are in the world
01:37:58
right now, like climate change being a
01:38:00
big one. Yeah. When people like Google
01:38:02
and Meta come out and be like, "No, but
01:38:04
the the biggest thing we need to solve
01:38:06
is like, how do we get your Google Docs
01:38:08
to write themselves and take your job?"
01:38:09
>> And and we're going to do the worst
01:38:12
thing America has done to the
01:38:14
environment in 50 years to accelerate
01:38:16
this. You know, they actually had a
01:38:18
very, very, very brief section about
01:38:20
weather models with AI.
01:38:22
>> Yeah,
01:38:23
>> that was super interesting because
01:38:25
weather models right now are just
01:38:26
throwing a ton of compute at a ton of
01:38:29
data from all over the country and
01:38:31
trying to decide what the weather's
01:38:33
going to be and how potentially
01:38:34
dangerous weather could happen in
01:38:36
certain places and trying to get as
01:38:37
accurate as possible. And just from all
01:38:39
the research I've been doing about
01:38:40
weather models, these AI models kind of
01:38:42
came out of nowhere and are immediately
01:38:45
surpassing these ancient models that
01:38:47
we've been using forever. And it's like,
01:38:48
holy this is like super useful for
01:38:50
like determining whether there's going
01:38:51
to be dangerous storms in these places.
01:38:53
>> Yeah.
01:38:53
>> And they just kind of breezed over that
01:38:55
and moved to the next thing
01:38:57
>> because there was a reannouncement.
01:38:58
>> Yeah.
01:38:59
>> Cuz they announced that
01:39:00
>> it's been around for like a year
01:39:01
basically since they started doing it.
01:39:04
But that's the type of stuff that could
01:39:06
make awesome headlines, super useful,
01:39:08
really great stuff. They also had like
01:39:10
this very brief thing at the beginning
01:39:12
where they're like, "We're going to cure
01:39:13
all disease." And then they just moved
01:39:14
on.
01:39:15
>> Yeah. Like wait wait wait wait wait wait
01:39:17
what? And I think the the proof that all
01:39:19
that is nonsense was in that in the the
01:39:22
OpenAI Tesla Elon court case when all in
01:39:25
like in all those internal emails that
01:39:27
got released. At no point were Sam and
01:39:30
Elon emailing each other like guys we're
01:39:32
about to cure all disease. Like we
01:39:35
should just put our differences aside
01:39:37
and cure all disease. No. The whole time
01:39:39
they're like this is going to take over
01:39:41
the world. I want to be the guy that's
01:39:42
the super billionaire running the world.
01:39:44
like like they don't even believe it's
01:39:46
going to cure all disease. It's just
01:39:47
something they they say.
01:39:49
>> But this is all this is what Google has
01:39:51
been doing for a decade. This is what
01:39:53
their AI teams were working on. It's
01:39:55
just that when ChatGBT came out and
01:39:56
everyone was like, "Well, we could get
01:39:58
money." They started turning it all into
01:39:59
consumer products and now we are where
01:40:00
we are.
01:40:01
>> Yeah.
01:40:01
>> But they've had these alpha fold tools.
01:40:04
They've had the weather models. They've
01:40:06
been doing this.
01:40:07
>> Yeah.
01:40:07
>> And it was just that they would work
01:40:09
very quietly in the background with the
01:40:11
people who it mattered for. Yeah.
01:40:13
>> And now they're trying to tell us all
01:40:14
what it meant for.
01:40:16
>> Yeah. As soon as Chad GBT launched, they
01:40:18
were like, "Oh, we have to make AI the
01:40:20
main consumer product as instead of just
01:40:22
using it as a background tool." Cuz
01:40:24
that's to your point, Marquez, about the
01:40:26
like the weather models just being
01:40:27
updated with AI features. What they used
01:40:30
to do and what Apple used to do would
01:40:31
just be like, "Look at this awesome tool
01:40:33
that you actually can have a use for.
01:40:35
It's powered by machine learning."
01:40:37
>> But they would actually tell you what
01:40:39
was useful about it. And now
01:40:40
everything's just a demo of something
01:40:42
that you could do but you would never
01:40:43
do.
01:40:43
>> They do it like the other way around
01:40:44
where they're like, "We have this new
01:40:45
feature. It's called AI BL."
01:40:47
>> Yeah.
01:40:48
>> And you're like, "Oh boy, another
01:40:49
thing."
01:40:49
>> Instead of, "Look at this cool thing. It
01:40:51
can do this. It's uses AI to do it."
01:40:55
>> Yeah. I think that's the problem with
01:40:57
like these these hot these words that
01:40:59
just get super hyped and there's like no
01:41:01
nuance in them, you know, like big data
01:41:03
or like internet of things or like they
01:41:05
just call it under this giant umbrella
01:41:06
and there's no nuance underneath it. Do
01:41:09
you know what word
01:41:10
is
01:41:12
does live up to the hype? My brain is
01:41:15
sorry.
01:41:16
>> Yeah.
01:41:16
>> Trivia is that
01:41:19
one.
01:41:19
>> You were ready though. You were ready.
01:41:22
>> You knew where I was going.
01:41:22
>> There's two music playing at once.
01:41:24
>> That was
01:41:26
hurt my brain.
01:41:28
>> Yeah. This has been a really rambly
01:41:29
episode.
01:41:30
>> It's really hot.
01:41:30
>> It was a rambly. I mean, anyway, let's
01:41:33
just get into trivia. Let's just get
01:41:35
into Adam.
01:41:36
>> I'm about to crash. Few more words could
01:41:37
have led to another.
01:41:38
>> This is going to be the hardest thing to
01:41:39
edit of all time.
01:41:40
>> Which of the following is not a real
01:41:43
Google project?
01:41:45
Firebase.
01:41:47
>> Bazal
01:41:48
angle. Apache beam.
01:41:51
>> How do you spell bazal?
01:41:53
>> B A Z A L. Basil.
01:41:57
>> Sure. Like art. Basil.
01:41:59
>> Well, that's e Oh,
01:42:00
>> that would be how like a tech company
01:42:02
making basil would spell it. That's an I
01:42:05
though.
01:42:06
>> I know. Well, no, it's I know I'm saying
01:42:08
things spelled different. It's also an
01:42:09
S.
01:42:10
>> Yeah, I think it would just be B A Z L.
01:42:14
>> Bzl actually, right?
01:42:16
>> You're right. Damn. Well, we'll think
01:42:18
about it. And just at the end like
01:42:20
usual, we'll be right back.
01:42:22
>> Damn.
01:42:31
>> Support for the show comes from Framer.
01:42:33
First impressions matter. a ton. That's
01:42:35
why having a top-of-the-line website to
01:42:36
catch potential customers is so vital.
01:42:38
So why not try Framer to help upgrade
01:42:40
your.com? Framer is an enterprisegrade
01:42:43
Noode website builder used by teams at
01:42:45
companies like Perplexity and Miro to
01:42:48
move faster. With real-time
01:42:49
collaboration and robust CMS with
01:42:51
everything you need for great SEO, not
01:42:53
to mention advanced analytics that
01:42:54
include integrated AB testing, your
01:42:56
designers and marketers are empowered to
01:42:58
build and maximize your.com from day
01:43:00
one. So whether you want to launch a new
01:43:02
website, test a few landing pages, or
01:43:04
migrate your full.com, Framer has
01:43:06
programs for startups, scaleups, and
01:43:08
large enterprises to make going from
01:43:10
idea to live site as easy and fast as
01:43:13
possible. So learn how you can get more
01:43:14
out of your.com framer framer specialist
01:43:17
or get started building for free today
01:43:19
at framer.com/wave
01:43:21
for 30% off a framer pro annual plan.
01:43:24
That's framer.com/wave
01:43:27
for 30% off. framer.com/wave. rules and
01:43:30
restrictions may apply. All right,
01:43:31
welcome back. We didn't talk about this
01:43:33
because Google didn't talk about it. Uh
01:43:35
they put it in a little Android
01:43:36
developer blog, but where OS 7
01:43:38
officially got announced, which for
01:43:40
those that don't know is the OS that
01:43:42
runs on your Google smartwatch. Um
01:43:44
there's some highle stuff here. There's
01:43:46
some stuff that's good. I'm just going
01:43:47
to run through it really quickly.
01:43:49
They're adding flexible and dynamic wear
01:43:51
widgets. So now they support new card
01:43:53
layouts which align very well for the
01:43:55
mobile format. So they look the same on
01:43:57
your phone and on your watch, which is
01:43:58
good. you now have live updates on your
01:44:00
watch. So if you know live updates like
01:44:03
Uber or whatever that can sort of show
01:44:05
you how close your car is, you can now
01:44:07
show that on your watch. It's great.
01:44:08
>> Or sports.
01:44:08
>> Or or sports.
01:44:10
>> Yeah. Like Arsenal winning the Premier
01:44:12
League.
01:44:14
>> Which is crazy that they didn't have
01:44:16
that on the watch before because it
01:44:18
seems like that'd be the best place to
01:44:19
look at it. But you know, whatever. Uh,
01:44:22
app functions allows developers to
01:44:25
integrate their apps with agents so
01:44:26
people can complete tasks using their
01:44:28
voice. That's helpful. Task automation.
01:44:31
Users can invoke and track task
01:44:33
automation like asking your phone to
01:44:34
order Door Dash for you. Now you can see
01:44:36
that on your watch. So previously you
01:44:38
would like ask your Pixel to order Door
01:44:40
Dash for you and you would see it
01:44:41
happening on your phone, but you
01:44:43
wouldn't be able to know what was
01:44:44
happening in real time on your watch.
01:44:45
Now you can see the steps it's taking on
01:44:47
on your watch to make sure if it's
01:44:49
accidentally ordering you like a car off
01:44:51
of Amazon instead of your burrito, you
01:44:53
can stop it, which is good. Uh workout
01:44:56
tracker. There is now a rich
01:44:57
standardized workout tracking experience
01:44:59
that includes heart rate monitoring,
01:45:00
media control, and other things. So now
01:45:03
if you're making a fitness app, there's
01:45:05
a standardized layout that you can use
01:45:07
to make sure that the watch app looks
01:45:09
good. So if you just made the app for
01:45:11
Android, you don't have to completely
01:45:13
redesign a separate watch app. you can
01:45:14
just use Google standardized one which
01:45:16
is cool. Uh there are enhanced system
01:45:18
media controls. So there is per app
01:45:20
media auto launch controls which means
01:45:22
that if you open Spotify like on your
01:45:25
phone the media controls with like
01:45:27
playback and like pause forward back up
01:45:30
down volume stuff like that Android.
01:45:32
>> Yeah. So now if you it's like if you
01:45:33
open YouTube music it doesn't show up on
01:45:35
your watch but if you open Spotify it
01:45:37
does show up in your watch because they
01:45:38
want to give you as much control as
01:45:40
possible I suppose. Uh there's also an
01:45:42
audio output switcher on the watch now.
01:45:44
So if you have your phone connected to
01:45:46
multiple sources, you say you wanted to
01:45:48
come out of the kitchen or you want to
01:45:49
come out of the bedroom, you can switch
01:45:50
it from your watch, which is helpful. Uh
01:45:53
and then they made they updated watch
01:45:55
face format, which is a way to make
01:45:57
watch faces. Now it's watch face format
01:45:59
5. There's enhanced alignment options,
01:46:01
auto size enhancements, blend modes,
01:46:03
stroke joints, and hierarchal settings.
01:46:06
That's about it. type.
01:46:10
>> I'm gonna use maybe one of those.
01:46:12
>> Yeah. Well,
01:46:13
>> it's so funny because that's something
01:46:14
that would have been so good visually on
01:46:16
a screen on a stage that they had for
01:46:18
two hours and decided
01:46:19
>> which is why I feel like they shouldn't
01:46:20
even have done the Android show. They
01:46:21
should have just made it half of IO and
01:46:24
then done the Gemini.
01:46:25
>> Yes. So then at an hour in we could all
01:46:27
>> and then they could talk about Gemini
01:46:30
for
01:46:30
>> their attention.
01:46:31
>> Yeah. Did you want to talk about this
01:46:33
camera thing?
01:46:33
>> I'm going to talk about it for five
01:46:34
seconds. I thought you might know more
01:46:35
about it, but
01:46:36
>> yeah, I was in the wilderness. I see.
01:46:39
>> Panasonic Lumix L10. I saw this because
01:46:41
on the X1006 subreddit, someone posted
01:46:44
is like, "Oh, is this the new X100
01:46:46
competitor?" Um, it seems like it's way
01:46:49
more of a a RX100 type competitor. But I
01:46:53
then looked at it and the coolest thing
01:46:55
about this is how the lens comes out.
01:46:58
It's like a lens cap that is built into
01:47:00
the camera and it opens up like a little
01:47:03
trig gate.
01:47:04
>> Yeah. It's got three
01:47:07
flaps. Yeah. It looks like the like
01:47:09
something in like Indiana Jones where
01:47:11
they had to like shove a relic into a
01:47:15
big stone door to open it and it opens
01:47:17
in these like three flaps and then this
01:47:18
zoom lens comes out of it. Anyways, it's
01:47:20
a small like compact point andoot
01:47:23
camera. It looks
01:47:25
>> pretty cool. Pedapixel has a whole
01:47:26
review on it if you actually care about
01:47:28
it, but the way it comes out looks
01:47:30
really cool.
01:47:31
>> Yeah,
01:47:31
>> that's all I have.
01:47:32
>> Hopefully you don't drop it while it's
01:47:33
open and then break one of the hinges
01:47:35
off.
01:47:35
>> Yeah, to be fair, dropping most cameras
01:47:36
in that op would be in a rough spot.
01:47:39
>> And to be fair, the flaps aren't that
01:47:40
important. They just kind of act as like
01:47:42
a dust protector.
01:47:43
>> Yeah, the gold color looks pretty sweet.
01:47:45
>> Yeah, it does.
01:47:46
>> Cool. All right, most important thing.
01:47:48
>> All right, so now that you guys have
01:47:50
made it this far into the podcast, I
01:47:51
know that you're real fans. you're you
01:47:54
made it this far past all the rambling
01:47:55
and all the stuff that we've talked
01:47:56
about. And so instead of hitting with a
01:47:58
code word in the comments, this is a
01:48:00
great time to share with you something
01:48:02
that we're really proud of that is maybe
01:48:03
the coolest thing we've ever done. So
01:48:06
you guys know the Marble Olympics, like
01:48:08
the Jealous Marble Run YouTube channel.
01:48:10
>> Is it jealous or jealous?
01:48:12
>> I only know about this because you guys
01:48:14
talk about it all the time.
01:48:16
>> You've probably It's come across your
01:48:17
feed at some point in the last decade.
01:48:19
They're kind of iconic.
01:48:20
>> I don't know if it has. They're they're
01:48:23
awesome. They have essentially like a
01:48:24
bunch of different events. It's like
01:48:26
marble Olympics. Like a bunch of
01:48:27
different events that they can run and
01:48:28
they all compete against each other.
01:48:30
>> Uh they
01:48:33
have sponsors.
01:48:35
>> Yeah,
01:48:35
>> these marble teams have sponsors.
01:48:37
>> I think you undersold it a little. these
01:48:39
these the marble events like there are
01:48:42
are sprints, there's hurdles, there's
01:48:44
like javelin, there's everything you
01:48:46
could there's marbula one which has
01:48:50
>> fullblown like attempted track
01:48:54
recreations in a marble run where each
01:48:56
marble is representing a different
01:48:58
driver in Formula 1 and they do
01:49:00
qualifying beforehand. like it's
01:49:03
completely out of this world with
01:49:05
incredible uh like they have the timing
01:49:08
metrics and everything down for
01:49:09
everyone. Anyways, yeah, it's awesome.
01:49:10
>> It's exactly what you're picturing.
01:49:12
>> Uh we we decided we'd love to get
01:49:15
involved and instead of just sponsoring
01:49:17
one team, we decided every channel that
01:49:20
we run should have a team so that we can
01:49:22
root for them. So, here on the Waveform
01:49:24
podcast, we do have a sponsored uh
01:49:27
marble team.
01:49:28
>> We do. Uh that would be well
01:49:30
specifically the Solar Flares are the
01:49:32
waveform team.
01:49:33
>> We're like an OG gels Marvel run team
01:49:35
like Solar Flares have been competing
01:49:36
for at least five years if not like way
01:49:38
longer. Yeah.
01:49:40
>> The Waveform, bro.
01:49:42
>> We need their rookie card.
01:49:44
>> Yeah, we are we are we are proudly
01:49:46
sponsoring that marble team here at
01:49:47
Waveform. But that also means the
01:49:48
Autofocus channel is also involved. They
01:49:51
are sponsoring the plasma team. The
01:49:53
studio channel is sponsoring the primary
01:49:56
team. I believe the studio channels team
01:49:58
has already been relegated. Yeah, they
01:49:59
got that fast. Yeah, they had a poor
01:50:01
some poor performances early. And the
01:50:03
MKBHD team, the blackjacks team is doing
01:50:05
really well.
01:50:08
>> So, you know, now that you've made it
01:50:10
this far, we highly recommend going over
01:50:13
and checking out what those marble teams
01:50:15
are up to and rooting for your favorite
01:50:17
one. Maybe you're going to for the
01:50:18
waveform team,
01:50:19
>> but you're clearly going to root for the
01:50:21
blackjack team. I'm going to give you
01:50:23
this microphone.
01:50:24
>> You're part of both of these channels,
01:50:25
Marquez. I know that the black needs to
01:50:28
be tested for performance.
01:50:30
>> So, well, it's I'm really proud team is.
01:50:33
Of course, Marquez is so good at
01:50:35
everything. Of course, he would pick one
01:50:37
of the teams and destroy all of us.
01:50:39
>> It's the Mad Black Marble Man.
01:50:40
>> One thing Marquez doesn't actually need
01:50:43
his competence to to do and run. He
01:50:45
still wins that. I'm very upset about
01:50:47
that.
01:50:47
>> And the season, it's a long season.
01:50:49
Okay. There's a lot of events.
01:50:50
>> You can still lose. Yeah, for sure.
01:50:52
>> Will you switch teams to our to our
01:50:53
collective team if we start winning?
01:50:55
>> Um, no. I'm going to stick with my team,
01:50:58
but you know, the Waveform team should
01:50:59
stick with their team, too. We we should
01:51:00
we should watch all these teams. We
01:51:02
should watch the events. Some stuff is
01:51:03
live, too. You can literally watch them
01:51:05
like as they happen. They're really
01:51:06
exciting. So, go check them out. We'll
01:51:08
link whatever we can below, so you can
01:51:10
go support and check out our teams.
01:51:12
>> Uh, yeah, marble runs. We will retweet
01:51:15
the first person who posts a picture
01:51:17
with the solar filler marble merch
01:51:19
watching waveform
01:51:21
>> the solar flare.
01:51:22
>> Is that fair? You can you can buy the
01:51:24
team marbles on
01:51:25
>> Okay. Or uh you can buy jerseys. I think
01:51:29
>> that's incredible.
01:51:29
>> Do you guys Do you guys know who the
01:51:31
marbles on the solar flare are?
01:51:33
>> I have the team roster pulled up.
01:51:34
>> The names of
01:51:35
>> Wait, they have names? I forgot names.
01:51:37
Yo, who's our roster? Coming from the
01:51:39
seaside city of Meteorine, the Solar
01:51:41
Flares participated in the Stardust
01:51:43
Classic and won the Herbbitamia in
01:51:46
Invitational, allowing them and the
01:51:48
Gliding Glaciers to be invited to
01:51:50
compete in Marble League 2021
01:51:52
qualifiers. We are represented by
01:51:56
Actually, Hold on. What am I doing?
01:51:59
Get the music on.
01:52:02
>> Why am I not even seeing? We are
01:52:04
represented by
01:52:07
Flare, Scorch, Radiance, Ember, and
01:52:11
Blaze.
01:52:14
>> Aren't you glad I have that button just
01:52:15
like
01:52:16
>> ready to go?
01:52:18
>> So, we will be big time watching our
01:52:21
marbles and supporting them through the
01:52:23
rest of the season. Uh because it's the
01:52:26
most exciting thing we've ever been a
01:52:27
part of. the the main screen in our
01:52:29
office is either our work calendar or
01:52:32
>> marble run Olympics going on the like 4
01:52:34
hour live streams of it and
01:52:36
>> it's true
01:52:36
>> or disc golf too.
01:52:38
>> Yeah. Yeah. Disc golf makes it on there.
01:52:40
That's definitely the bottom.
01:52:42
>> Yeah.
01:52:43
>> So anyway, thank you guys for watching
01:52:45
and listening this week. I think it's
01:52:46
time for that last segment that we
01:52:48
always do.
01:52:49
>> Trivia.
01:52:50
>> Trivia.
01:52:52
>> Do you have a whiteboard? Oh, you do?
01:52:53
>> Yeah. Guys, I have before me four
01:52:55
women's names, specifically Rachel,
01:52:58
Mimi, Robin, and Shakira.
01:53:01
What line of smartphones had all four of
01:53:03
these as code names?
01:53:11
H
01:53:18
>> um
01:53:21
>> you are about to experience things that
01:53:23
you can't believe.
01:53:29
>> Getting carried away over here.
01:53:30
>> Behind the trivia music is crazy.
01:53:33
>> There's so many good ones. David, we put
01:53:34
this one in the soundboard.
01:53:41
>> That honestly sounds exactly like my
01:53:43
flex printer, right?
01:53:44
>> No, no, it's a scanner.
01:53:46
>> What'd you put?
01:53:46
>> Yeah.
01:53:48
>> Okay.
01:53:49
>> All right.
01:53:50
>> Marquez, you want to read first?
01:53:52
>> I said Samsung's A series, but I also
01:53:54
wanted to point out that there is
01:53:55
literally a phone call. The Robin,
01:53:58
that's good.
01:53:59
>> I was thinking that, too.
01:54:00
>> Andrew David, do you want to read your
01:54:01
answers at the exact same time? Ours are
01:54:03
not the same answer.
01:54:03
>> Yes, they are.
01:54:05
>> Well, let's
01:54:06
>> say with the same.
01:54:07
>> You said a lineup of phone, correct?
01:54:09
>> Hold on, hold on, hold on, hold on. You
01:54:10
said a lineup of phone. If he said
01:54:12
>> Oh my god, you've been back.
01:54:13
>> If he said Samsung, if he said Samsung
01:54:16
and it was the A series, would you have
01:54:18
counted that?
01:54:19
>> Does Sony even make
01:54:21
>> Xperia? They make you I think
01:54:24
>> you just wrote Xperia. They have Xperia
01:54:26
1. They also have Xperia 5. They have
01:54:27
Xperia 10. What? What were the things? I
01:54:30
actually was debating putting one or
01:54:32
five,
01:54:32
>> but you didn't.
01:54:33
>> Are they both right?
01:54:34
>> Well, you just put Sony.
01:54:35
>> Is one of them right?
01:54:36
>> No, they're they're
01:54:38
>> Guys, what even is this? Come on, man.
01:54:40
You've been back
01:54:40
>> because I was originally thinking of an
01:54:41
LGR and I was like, do I put V series
01:54:44
>> Sony Xperia? That's what I meant.
01:54:46
>> How how how specific are you? Okay,
01:54:48
look. You haven't told me the answer
01:54:49
yet. What are they? What are the four?
01:54:52
>> Those are all Sony Xperia phones.
01:54:54
>> Xperia what?
01:54:55
>> Xperia. So these are all these are the
01:54:57
first four Xperia phones that ran
01:54:59
Android after they switched from Windows
01:55:02
phone.
01:55:02
>> So wasn't what were the name? Yeah.
01:55:04
>> The Xperia X10A, the Xperia U20A, the
01:55:09
Xperia E10A, and the Xperia E15. Oh,
01:55:12
wait. Those are the the model. Excuse
01:55:14
me. The Xperia X10, the X10 Mini Pro,
01:55:17
the X10 Mini, and the Xperia X8.
01:55:20
>> Xperia X. So,
01:55:21
>> I could give neither of you the point
01:55:23
because neither you put Xperia X, but I
01:55:25
think
01:55:26
>> what's the score?
01:55:26
>> Is there a Sony phone that's not an
01:55:28
Xperia Xeria?
01:55:29
>> They want to be specific. So, they got
01:55:31
it all wrong.
01:55:32
>> The question was what line and it's the
01:55:34
Xperia line.
01:55:35
>> Yeah, but nobody would would have been
01:55:36
>> I just feel like if you said that line
01:55:38
or that what company,
01:55:40
>> but Sony only makes Xperia phones.
01:55:42
>> I think they have more than one line.
01:55:43
>> Ellis was going to get both of you.
01:55:45
>> Put the other lines.
01:55:47
>> So, do we both get it wrong?
01:55:48
>> I don't know. Sony doesn't make
01:55:49
non-experiences.
01:55:51
>> No.
01:55:51
>> So, does Sony make non experience?
01:55:52
>> I don't want to experience.
01:55:54
>> It's making me look bad, but I don't
01:55:56
know. That feels like a very vague
01:55:57
answer.
01:55:59
>> But there's only one I'm just saying.
01:56:02
They're all experience.
01:56:03
>> I don't know what to do. I don't I don't
01:56:04
know what to do here. Okay, here's
01:56:06
here's the thing. If I give Andrew the
01:56:09
point, you know, then he's tied with
01:56:10
Marquez. We get like this nice
01:56:12
competitive thing.
01:56:13
>> Oh,
01:56:13
>> and then if I give David the point, he's
01:56:15
even farther ahead of everyone else. I
01:56:17
think that should be taken.
01:56:18
>> I think we should both get points.
01:56:21
>> Yeah. Yeah. Neither of you get a point
01:56:22
right now. In the comments, let us know
01:56:26
where David Andrew both said that I
01:56:28
lost.
01:56:30
>> Get the point.
01:56:31
>> What does What does Gemini say?
01:56:33
>> It was so implied.
01:56:34
>> We'll let the comments section figure
01:56:35
this one out.
01:56:35
>> It was so implied.
01:56:36
>> Wow, guys. All right. Well, since no
01:56:38
points have been awarded,
01:56:39
>> be so bring Mariah back.
01:56:42
>> Top number one comment. Easy. Andrew,
01:56:45
you've officially initiated my grudge.
01:56:47
>> Right. I would like to kick off hour 39
01:56:50
of this podcast episode with an update
01:56:52
to the score. Uh Andrew, you are
01:56:55
carrying the one with 25 points. Maybe
01:56:58
26. We'll see. Uh Marquez, you are just
01:57:00
ahead of him with 26 actual real points.
01:57:04
And David, you are still far in the lead
01:57:07
with 29 points.
01:57:09
>> There's a point where I was like double
01:57:11
both of their scores put together. No,
01:57:12
you did start this season with a huge
01:57:14
hugely. We're also so overdue for an
01:57:17
extravaganza.
01:57:17
>> Yeah, like significantly.
01:57:19
>> Question number two, which of the
01:57:20
following is not a real Google project?
01:57:22
A. Firebase, B. Buzzal, C. Angle, or C.
01:57:28
Apache, or D, sorry, Apache Beam.
01:57:34
H.
01:57:38
What do you guys think?
01:57:42
Apache beam more like
01:57:46
it's so hot in here.
01:57:51
>> Flip them and read. What do you got?
01:57:54
>> Wow. Okay, I'm different again from the
01:57:56
other two. I said B. Bazal
01:58:00
>> wrong. That is a real answer.
01:58:02
>> Spelled it wrong.
01:58:03
>> I think the spelling is
01:58:04
>> it's B A Z E L.
01:58:07
>> Like basil.
01:58:08
>> That's not how you spelled it before,
01:58:09
was it? That's not how you pronounce
01:58:11
that.
01:58:11
>> Oh, I don't know.
01:58:12
>> Basel.
01:58:12
>> You spelled it b a za.
01:58:15
>> Bazelle. Bazelle. It's a
01:58:16
>> tomato. Tomato.
01:58:17
>> It's their own.
01:58:18
>> Sony Experia.
01:58:19
>> Yeah.
01:58:20
>> Their what tool?
01:58:21
>> It's a build tool that has built-in
01:58:23
support for building both client and
01:58:24
server software.
01:58:26
>> And then we said I don't I don't know,
01:58:29
bro. I don't know. This is on their
01:58:30
page. Google open source. All right.
01:58:32
>> It's called Basil.
01:58:33
>> Basil.
01:58:33
>> Is that how you pronounce it?
01:58:34
>> Yeah. You said bazol. I thought that was
01:58:36
real.
01:58:37
>> Baz. Bazal. It's literally called basil.
01:58:39
>> Well, it was Well, what Adam said wasn't
01:58:42
real.
01:58:43
>> Exactly. That's why I should get the
01:58:45
point.
01:58:46
>> No, but it was real. It's a real thing.
01:58:48
>> Bazal is not real.
01:58:49
>> Well, I don't know. I've never heard
01:58:51
this spoken out loud. It could very well
01:58:53
be Bazal.
01:58:54
>> It's I Googled it.
01:58:56
>> And it's pronounced Bazal.
01:58:57
>> No, it's called Basil,
01:58:59
>> but it's not spelled like Basil.
01:59:00
>> Not real. It's a tech company.
01:59:03
>> Here, this is how Google pronounces it.
01:59:06
>> Basil. What does Google know?
01:59:10
>> Andrew, David, what did you guys put?
01:59:12
>> C or angle.
01:59:15
>> I forgot what it was for us. And I put
01:59:18
both to be specific.
01:59:20
>> That is
01:59:20
>> You're going to have to be specific.
01:59:22
>> And I put an angle.
01:59:24
>> All right. Well, this is all going to be
01:59:26
determined in the comments because I'm
01:59:28
not adding a point on the screen until
01:59:31
>> you guys tell me whether Marquez should
01:59:33
get that point or not.
01:59:34
>> I feel like I've been robbed again. I
01:59:36
feel like Bazal is clearly not real. And
01:59:39
I should
01:59:39
>> Well, but didn't you just say that you
01:59:41
the audience will decide.
01:59:42
>> You said, "Tell me which one's not real.
01:59:44
Google Bazall." I said, "That one's not
01:59:45
real."
01:59:46
>> But one of them is very obviously not
01:59:48
real.
01:59:48
>> Yeah, but I was right. There's two
01:59:50
correct answers.
01:59:52
>> We'll see.
01:59:53
>> We'll see.
01:59:53
>> Commenters,
01:59:54
>> commenters, do your do your thing. Do
01:59:56
your magic. Uh, wow. Okay. There's
01:59:58
probably six minutes left on this card
02:00:00
and it's 1,000 degrees in here. So,
02:00:02
thanks for watching, thanks for
02:00:03
listening, and thanks for subscribing.
02:00:05
Stay tuned for more of your regularly
02:00:06
scheduled programming, but also some
02:00:07
bonus programming coming up soon because
02:00:09
we're just we're just that committed to
02:00:11
you guys. And if you made it this far,
02:00:14
comment bazal.
02:00:17
>> Comment bring Mariah back.
02:00:19
>> They already will.
02:00:20
>> They already They already did.
02:00:22
>> See you guys next time. Peace.
02:00:24
>> Waveform is produced by Vox Media
02:00:26
Podcast Network. We're
02:00:27
>> now Wform was produced by Vox Media.
02:00:30
>> Nope.
02:00:30
>> You're doing great, sweetie.
02:00:31
>> Waveform was produced by Adam Molina and
02:00:33
Ellis Roven. We're partner of Vax Media
02:00:35
Podcast Network and try out music was
02:00:36
created by Vans Still.
02:00:45
>> All right. How long are you guys talking
02:00:47
about the Fuji? because I got some
02:00:48
marbles to talk
