00:00:00
So much of what's happening today in the
00:00:02
AI industry is extremely inhumane.
00:00:04
>> But this is me playing devil's advocate.
00:00:06
And logically, it could be the case that
00:00:08
the civilization that accelerate their
00:00:10
research with AI is going to be the
00:00:12
superior civilization.
00:00:13
>> No, it's not. This is a prediction that
00:00:14
you're making, right?
00:00:15
>> Making Zuckerberg's making.
00:00:18
>> And do you know what the common feature
00:00:19
of all of them is? They profit
00:00:20
enormously off of this myth. You know, I
00:00:22
have all these internal documents
00:00:24
showing that they're purposely trying to
00:00:26
create that feeling within the public so
00:00:28
that they can extract and exploit and
00:00:30
extract and exploit. So, what do we do
00:00:32
about it?
00:00:32
>> We need to break up the empires of AI.
00:00:35
>> You know, I've been covering the tech
00:00:36
industry for over 8 years, interviewed
00:00:38
over 250 people, including former or
00:00:40
current OpenAI employees and executives.
00:00:42
And I can tell you that there are many
00:00:44
parallels between the empires of AI and
00:00:46
the empires of old, right? like Lelay
00:00:48
claimed the intellectual property of
00:00:49
artists, writers, and creators in the
00:00:50
pursuit of training these models.
00:00:52
Second, they exploit an extraordinary
00:00:53
amount of labor, which breaks the career
00:00:55
ladder because someone gets laid off and
00:00:57
then they work to train the models on
00:01:00
the very job that they were just laid
00:01:01
off in, which will then perpetuate more
00:01:03
layoffs if that model then develops that
00:01:05
skill. And when they talk about that
00:01:07
there's going to be some new jobs
00:01:08
created that we can't even imagine, a
00:01:10
lot of the jobs that are created are way
00:01:12
worse than the jobs that were there. And
00:01:14
then there's the environmental and
00:01:15
public health crisis that these
00:01:17
companies have created and how they're
00:01:20
able to also spend hundreds of millions
00:01:21
to try and kill every possible piece of
00:01:23
legislation that gets in their way and
00:01:25
will censor researchers that are
00:01:27
inconvenient to the empire's agenda. But
00:01:30
what I'm saying is not that these
00:01:32
technologies don't have utility. It's
00:01:34
that the production of these
00:01:35
technologies right now is exacting a lot
00:01:37
of harm on people. But we have research
00:01:39
that shows that the very same
00:01:42
capabilities could be developed in a
00:01:44
different way that doesn't have all of
00:01:46
these unintended consequences. So let's
00:01:48
talk about all of that.
00:01:53
This is super interesting to me. My team
00:01:54
given me this report to show me how many
00:01:56
of you that watch this show subscribe.
00:01:57
And some of you have told us according
00:01:59
to this that you are unsubscribed from
00:02:01
the channel randomly. So favor to ask
00:02:03
all of you. Please could you check right
00:02:04
now if you've hit the subscribe button
00:02:06
if you are a regular viewer of the show
00:02:07
and you like what we do here. We're
00:02:08
approaching quite a significant landmark
00:02:10
on this show in terms of a subscriber
00:02:12
number. So if there was one simple free
00:02:14
thing that you could do to help us, my
00:02:16
team, everyone here to keep this show
00:02:18
free, to keep it improving year over
00:02:20
year and week over week, it is just to
00:02:22
hit that subscribe button and to double
00:02:23
check if you've hit it. Only thing I'll
00:02:24
ever ask of you, do we have a deal? If
00:02:27
you do it, I'll tell you what I'll do.
00:02:28
I'll make sure every single week, every
00:02:31
single month, we fight harder and harder
00:02:32
and harder and harder to bring you the
00:02:33
guests and conversations that you want
00:02:34
to hear. I've stayed true to that
00:02:36
promise since the very beginning of the
00:02:37
D of Sio and I will not let you down.
00:02:40
Please help us. Really appreciate it.
00:02:42
Let's get on with the show.
00:02:47
Karen, how you've written this book in
00:02:50
front of me here called Empire of AI:
00:02:52
Dreams and Nightmares in Sam Alman's
00:02:55
Open AI. I guess my first question is
00:02:58
what is the research and the journey you
00:03:00
went on in order to write this book
00:03:02
we're going to talk about and the
00:03:03
subjects within it today
00:03:04
>> I took a strange route into journalism I
00:03:07
studied mechanical engineering at MIT
00:03:09
and so when I graduated I moved to San
00:03:11
Francisco I joined a tech startup I
00:03:13
became part of Silicon Valley and I
00:03:16
basically received an education in what
00:03:18
Silicon Valley is about because a few
00:03:19
months into joining a very missiondriven
00:03:21
startup that was focused on building
00:03:23
technologies that would help facilitate
00:03:25
the fight against climate change. The
00:03:27
board fired the CEO because the company
00:03:29
was not profitable. And this was in
00:03:32
hindsight a very pivotal moment for me
00:03:35
because I thought if this hub is
00:03:37
ultimately geared towards building
00:03:40
profitable technologies and many of the
00:03:43
problems in the world that I think need
00:03:45
solved are not profitable problems like
00:03:47
climate change. Then what are we
00:03:50
actually doing here? like what how did
00:03:52
we get to a point where innovation is
00:03:54
not actually necessarily working in the
00:03:57
public benefit and sometimes even
00:03:58
undermining the public benefit in
00:04:00
pursuit of profit. In that moment, I had
00:04:03
a bit of a crisis where I thought, well,
00:04:06
I just spent 4 years trying to set
00:04:10
myself up for this career that I now
00:04:11
don't think I am cut out for. And I
00:04:16
thought, well, I might as well just try
00:04:18
something totally different. I've always
00:04:20
liked writing and that's how after 2
00:04:22
years I landed at a role at MIT
00:04:26
technology review covering AI full-time
00:04:28
and that gave me a space to then explore
00:04:31
all of these questions of who gets to
00:04:33
decide what technologies we build how
00:04:35
does money and ideology also drive the
00:04:38
production of those technologies and how
00:04:40
do we ultimately make sure that we
00:04:42
actually reimagine the innovation
00:04:44
ecosystem to work for a broad base of
00:04:48
people all around the world. And so that
00:04:51
is kind of how I then set off on this
00:04:53
journey of ultimately writing a book. I
00:04:56
didn't realize that I was working
00:04:58
towards writing a book, but starting in
00:05:01
2018 when I took that job was
00:05:04
essentially the moment in which I began
00:05:06
researching the story that I I document
00:05:08
in it.
00:05:09
>> A very timely time to start working in
00:05:11
artificial intelligence. For anyone that
00:05:12
doesn't know, this is pre OpenAI chat
00:05:14
GPT launch moment that shook the world.
00:05:19
But in writing this book, you
00:05:20
interviewed a lot of people and went to
00:05:21
a lot of places. Can you give me a
00:05:22
flavor of how many people you've
00:05:24
interviewed, where it's taken you around
00:05:26
the world, etc.
00:05:27
>> I interviewed over 250 people. So over
00:05:29
300 interviews, over 90 of those people
00:05:32
were former or current OpenAI employees
00:05:35
and executives. So the book covers the
00:05:38
inside story of opening eyes's first
00:05:40
decade and how it ultimately got to
00:05:42
where it is today. But I didn't want to
00:05:45
write a corporate book. I felt very
00:05:47
strongly that in order to help people
00:05:49
understand the impact of the AI
00:05:52
industry, we would also have to travel
00:05:54
well beyond Silicon Valley. These
00:05:56
companies tell us that AI is going to
00:05:58
benefit everyone and that's their
00:05:59
mission. But you really start to see
00:06:02
that rhetoric break down when you go to
00:06:05
the places that look nothing like
00:06:07
Silicon Valley, that speak nothing like
00:06:09
Silicon Valley, and that have a history
00:06:11
and culture that are fundamentally
00:06:12
different as well. And that's where you
00:06:14
start to really understand the true
00:06:17
reality of how this industry is
00:06:21
unfolding around us.
00:06:22
>> Karen, I often try and steer
00:06:24
conversations, but in this situation, I
00:06:26
feel like it's probably my
00:06:27
responsibility to follow. So with that
00:06:30
in mind, I'm going to ask you where does
00:06:32
this journey begin and where should we
00:06:33
be starting if we're talking about the
00:06:34
subjects of empire of AI, AI generally
00:06:38
artificial intelligence and also I'd say
00:06:40
one thing I'm really keen to do in this
00:06:42
conversation which is I often see in
00:06:43
conversations is left out is let's
00:06:46
assume that our viewers know nothing
00:06:47
about AI.
00:06:48
>> Yeah. So they don't know what scaling
00:06:50
laws are or GPUs or comput or whatever
00:06:52
and let's try and keep this as simple as
00:06:54
we possibly can in terms of language or
00:06:57
explain all the complicated language so
00:06:59
that we can bring as much people with us
00:07:00
as we possibly can.
00:07:01
>> Yes.
00:07:02
>> Where should we start?
00:07:03
>> I think we should start with when AI
00:07:06
started as a field. So this was back in
00:07:09
1956
00:07:11
and there were a group of scientists
00:07:13
that gathered at Dartmouth University to
00:07:15
start a new discipline, a scientific
00:07:17
discipline to try and chase an ambition.
00:07:20
And specifically an assistant professor
00:07:22
at Dartmouth University, John McCarthy
00:07:24
decided to name this discipline
00:07:25
artificial intelligence.
00:07:28
This was not the first name that he
00:07:29
tried. The previous year he tried to
00:07:32
name it Automata Studies. And the reason
00:07:34
why some of his colleagues were
00:07:36
concerned about this name was because it
00:07:38
pegged the idea of this discipline to
00:07:41
recreating human intelligence. And back
00:07:44
then, as is true today, we have no
00:07:47
scientific consensus around what human
00:07:50
intelligence is. There's no definition
00:07:52
from psychology, biology, neurology. And
00:07:55
in fact, every attempt in history to
00:07:59
quantify and rank human intelligence has
00:08:02
been driven by nefarious motives. It's
00:08:05
been driven by a desire to prove
00:08:09
scientifically that certain groups of
00:08:10
people are inferior to other groups of
00:08:13
people. There are no goalposts for this
00:08:17
field and there are no goalposts for the
00:08:19
industry when they say that they are
00:08:21
ultimately trying to recreate AI systems
00:08:24
that would be as smart as humans. How do
00:08:26
we even define what that means? And when
00:08:29
are we going to get there if we don't
00:08:31
know how to define the destination? And
00:08:34
what that effectively means is that
00:08:37
these companies can just use the term
00:08:39
artificial general intelligence which is
00:08:41
now the term to refer to this ambitious
00:08:44
um goal to recreate human intelligence.
00:08:47
They can use it however they want to and
00:08:49
they can define and redefine it based on
00:08:51
what is convenient for them. So in
00:08:53
OpenAI's history, it has defined and
00:08:55
redefined it many times. When Sam Alman
00:08:57
is talking with Congress, AGI is a
00:09:00
system that's going to cure cancer,
00:09:02
solve climate change, cure poverty. When
00:09:05
he's talking with consumers that he's
00:09:07
trying to sell his products to, it's the
00:09:09
most amazing digital assistant that
00:09:11
you're ever going to have. When he was
00:09:14
talking with Microsoft, you know, in the
00:09:16
deal that OpenAI and Microsoft struck
00:09:18
where Microsoft invested in the company,
00:09:21
it was defined as a system that will
00:09:23
generate hundred billion of revenue. And
00:09:26
on OpenAI's own website, they define it
00:09:28
as highly autonomous systems that
00:09:30
outperform humans in most economically
00:09:33
valuable work. This is like not a
00:09:36
coherent vision of one technology. These
00:09:39
are very different definitions that are
00:09:41
spoken out loud to the audience that
00:09:44
needs to be mobilized to ward off
00:09:47
regulation or get more consumer buy in
00:09:50
into the the industry's quest or to get
00:09:54
more capital more resources for
00:09:56
continuing on this journey with
00:09:58
ambiguous definitions. I mean, speaking
00:10:01
about different definitions through
00:10:02
time, in 2015, in a blog post that Sam
00:10:06
Waltman wrote before open air was
00:10:07
officially announced, he explicitly
00:10:10
outlined the existential risk by saying,
00:10:12
"Development of superhuman machine
00:10:14
intelligence is probably the greatest
00:10:16
threat to the continued existence of
00:10:18
humanity. There are other threats that I
00:10:20
think are more certain to happen, for
00:10:21
example, an engineered virus, but AI is
00:10:24
probably the most likely way to destroy
00:10:27
everything
00:10:28
>> in general." When Alman is writing for
00:10:31
the public or speaking for the public,
00:10:33
he does not just have the public as the
00:10:35
audience in mind, there are other people
00:10:38
that he is trying to motivate or
00:10:40
mobilize when he says these things. And
00:10:43
in that particular moment, Alman was
00:10:46
trying to convince Elon Musk to join him
00:10:48
on co-founding OpenAI. And Musk in
00:10:52
particular was spending all of his time
00:10:55
sounding the alarm on what he saw as a
00:10:58
huge existential threat that AI could
00:11:00
pose. And so in that blog post, if you
00:11:03
look at the the language that Alman uses
00:11:05
side by side with the language that Musk
00:11:07
was using at the time, it mirrors all
00:11:10
the things that Musk was saying
00:11:11
>> identical. I mean, 10 years ago, Musk
00:11:13
was going on podcast saying, tweeting,
00:11:16
whatever, that the greatest existential
00:11:18
risk to humanity was AI.
00:11:19
>> Yeah. And so you know like his
00:11:21
parenthetical there are other things
00:11:23
that we that might actually be more
00:11:26
likely to happen like engineered
00:11:27
viruses. It's because up until then
00:11:29
Alman had been talking just about
00:11:32
engineered viruses. And so now that he
00:11:35
needs to pivot to speak to an audience
00:11:37
of one to Musk. He needs to kind of
00:11:40
resolve the contradiction between what
00:11:42
he's now elevating as his new central
00:11:45
fear to be the same as Musk's new
00:11:47
central fear with what he had previously
00:11:49
been saying. So that's why he's like I
00:11:51
think this is now even though before I
00:11:54
said this
00:11:55
>> and are you saying that Sam Alman
00:11:57
manipulated Musk because Elon did end up
00:12:01
donating a huge amount of money to um
00:12:03
open AAI and co-founding it I believe
00:12:06
with Sam Alman. Elon Musk did end up
00:12:07
co-ounding it with Altman. And certainly
00:12:09
from Musk's perspective, he does feel
00:12:12
manipulated because he feels like Alman
00:12:16
was engineering his language in a way
00:12:20
that would make Musk trust him as a a
00:12:24
partner in this endeavor. And of course
00:12:27
then Musk is leaves. Um and through some
00:12:31
of the documents that came out during
00:12:32
the the lawsuit that Musk and Altman are
00:12:35
engaged in now, it has become clear that
00:12:38
there was a degree to which Musk was
00:12:40
actually muscled out a little bit. And
00:12:43
so that's why he's left with this
00:12:47
very intense personal vendetta against
00:12:48
Altman, saying that somehow Alman
00:12:51
tricked him into being part of this. So
00:12:54
in in 2015, Sam Alman is writing these
00:12:56
blog posts saying this is, you know, one
00:12:58
of the greatest existential threats. At
00:12:59
the same time, in 2015, Musk is doing
00:13:02
some very famous speeches at the time at
00:13:04
MIT. He said that AI was the biggest
00:13:06
existential threat and compared
00:13:08
developing AI to summoning the demon.
00:13:11
And what you're saying here is you're
00:13:12
saying that Samman was just mirroring
00:13:14
the language that Elon was using to get
00:13:16
Elon involved in open open AAI. And
00:13:18
later it appears and again there's a
00:13:20
legal case taking place now that Sam
00:13:23
might have muscled Elon out in some
00:13:24
capacity.
00:13:25
>> Yeah. So we know from the lawsuit and
00:13:27
the documents that have come out in the
00:13:28
lawsuit that Ilia Sgver who is the chief
00:13:33
scientist of OpenAI at the time and Greg
00:13:35
Brockman chief technology officer at the
00:13:37
time when they were deciding whether or
00:13:40
not to maintain OpenAI as a nonprofit
00:13:43
because it was originally founded as a
00:13:44
nonprofit. They decided okay we need to
00:13:46
create a for-profit entity but the
00:13:47
question was who should be the CEO of
00:13:49
this for-profit entity. Should it be
00:13:50
Musk or should it be Alman? because it's
00:13:52
they were the two co-chairmen of the
00:13:54
nonprofit. And in the emails, it became
00:13:58
clear that Ilia and Greg first chose
00:14:01
Musk to be the CEO.
00:14:05
But through my reporting, I discovered
00:14:08
that Altman then appealed personally to
00:14:11
Greg Brockman, who was a friend of his
00:14:13
that they had known, they had known each
00:14:15
other for many years through the Silicon
00:14:16
Valley scene, and said, "Don't you think
00:14:20
that it would be a little bit dangerous
00:14:22
to have Musk be the CEO of this company,
00:14:26
this new for-profit entity, because, you
00:14:28
know, he's a famous guy. He has a lot of
00:14:31
pressures in the world. He could be
00:14:33
threatened. He could act erratically. He
00:14:36
could be unpredictable. And do we really
00:14:38
want a technology that could be super
00:14:41
powerful in the future to end up in the
00:14:43
hands of this man? And that convinced
00:14:46
Greg and Greg then convinced Ilia, you
00:14:49
know, I think there's a point here. Do
00:14:52
we really want to give this much power
00:14:54
to Musk? And that is why Musk then
00:14:57
leaves because then they the two switch
00:15:00
their allegiances. They say, "Actually,
00:15:02
we want Altman to be the CEO." And then
00:15:04
Musk is like, "If I'm not CEO, I'm out."
00:15:06
>> So, it sounds like Sam again managed to
00:15:08
persuade someone to do something.
00:15:10
>> Mhm.
00:15:12
>> I guess this begs the question, what do
00:15:14
you think of Sam Orman?
00:15:17
>> I think he's a very controversial
00:15:18
figure.
00:15:19
>> You did an interesting pause. It's a
00:15:22
pause where someone tries to select
00:15:24
their words. Well, this is this is this
00:15:27
is what's so interesting
00:15:30
about those interviews is people are
00:15:33
extremely polarized on Alman there. No
00:15:36
one has in between feelings about him.
00:15:39
Either they think he's the greatest tech
00:15:41
leader of this generation akin to the
00:15:43
Steve Jobs of the modern era or they
00:15:46
think that he's really manipulative and
00:15:49
an abuser and a liar. And what I
00:15:53
realized because I interviewed so many
00:15:55
people is it really comes down to what
00:15:58
that person's vision of the future is
00:16:00
and what their goals are. So if you
00:16:04
align with Altman's vision of the
00:16:06
future, you're going to think he's the
00:16:08
greatest asset ever to have on your side
00:16:10
because this man is really persuasive.
00:16:12
He's incredible at telling stories. He's
00:16:14
incredible at mobilizing capital, at
00:16:16
recruiting talent, at getting all the
00:16:18
inputs that you need to then make that
00:16:20
future happen. But if you don't agree
00:16:23
with his vision of the future, then you
00:16:26
begin to feel like you're being
00:16:28
manipulated by him to support his vision
00:16:33
even if you fundamentally don't agree
00:16:34
with it. And this is the story
00:16:36
especially of Daria Amade, CEO of
00:16:39
Enthropic, who was originally an
00:16:41
executive at OpenAI. So for people that
00:16:43
don't know, Dario now runs anthropic
00:16:45
which is the maker of Claude. A lot of
00:16:47
people probably are more familiar with
00:16:48
Claude.
00:16:49
>> Yeah. And it's one of the biggest
00:16:51
competitors to OpenAI.
00:16:53
And Amade at the time when he was an ex
00:16:57
executive at OpenAI,
00:16:59
he thought that Alman was on the same
00:17:03
page with him and then over time began
00:17:06
to feel that Altman was actually on
00:17:08
exactly the opposite page of him and
00:17:11
felt that Altman had used Amade's
00:17:15
intelligence, capabilities, skills to
00:17:19
build things and bring about a vision of
00:17:22
the future that he actually
00:17:23
fundamentally didn't agree with. And so
00:17:25
that's why people end up with this bad
00:17:28
taste in their mouths. And so, you know,
00:17:30
I've been covering the tech industry for
00:17:33
over eight years and covered many
00:17:35
companies. I've covered Meta, Google,
00:17:36
Microsoft in addition to Open AI. and
00:17:39
OpenAI and Altman is it's the only
00:17:43
figure that I've seen this degree of
00:17:45
polarization with where people cannot
00:17:48
decide
00:17:50
whether he's the greatest or the worst.
00:17:53
>> You mentioned Dario there and I found it
00:17:56
really what I found really interesting
00:17:57
is to look at how people's quotes evolve
00:17:59
over time with their incentives. So I
00:18:01
was looking at all of the all of the
00:18:03
things they've said on the record on
00:18:04
podcasts in their blog post to see how
00:18:06
it's evolved over time and Dario who was
00:18:08
the former VP of research open AAI and
00:18:11
has now moved on to enthropic who are
00:18:13
taking a slightly different approach to
00:18:14
developing AI said back in 2017 while he
00:18:17
was still at open AI that this is a
00:18:20
quote I think at the extreme end is the
00:18:22
Nick Bostonramm style of fear that an
00:18:24
AGI could destroy humanity. I can't see
00:18:27
any reason in principle why that
00:18:29
couldn't happen. My chance that
00:18:31
something goes really quite
00:18:33
catastrophically wrong on the scale of
00:18:34
human civilization
00:18:36
might be somewhere between 10% and 25%.
00:18:40
And also you mentioned Ilia who was a
00:18:43
co-founder of OpenAI and then left. I
00:18:45
guess the first question I'd ask is why
00:18:47
did I leave?
00:18:49
>> It's a great question.
00:18:52
So he was instrumental in trying to get
00:18:54
Sam Alman fired and he's another one of
00:18:58
the people who over time began to feel
00:19:00
like he was being manipulated by Alman
00:19:03
towards contributing something that he
00:19:05
didn't believe in. And for
00:19:07
>> you know
00:19:07
>> because I interviewed a lot of people
00:19:09
Ilia in particular had
00:19:12
two pillars that he cared about deeply.
00:19:16
One is making sure we get to so-called
00:19:19
AGI and the other is making sure that we
00:19:22
get to it safely. And he felt that
00:19:25
Altman was actively undermining both
00:19:28
things. He felt that Alman was creating
00:19:31
a very chaotic environment within the
00:19:33
company where he was pitting teams
00:19:35
against each other where he was telling
00:19:37
different things to different people.
00:19:39
>> Have you ever spoken to him?
00:19:40
>> I have. So, so I interviewed him in 2019
00:19:43
for a profile that I did of OpenAI um
00:19:47
for MIT Technology Review
00:19:48
>> and back in 2019, he has a quote where
00:19:51
he says, "The future's going to be good
00:19:52
for AIs regardless. It would be nice if
00:19:54
it was also good for humans as well.
00:19:56
It's not that it's going to actively
00:19:58
hate humans or want to harm them, but
00:19:59
it's just going to be so powerful. And I
00:20:01
think a good analogy would be the way
00:20:02
that humans treat animals. It's not that
00:20:04
we hate animals. I think humans love
00:20:06
animals, and I have a lot of affection
00:20:08
for them. But when the time comes to
00:20:10
build a highway between two cities, we
00:20:11
are not asking the animals for
00:20:13
permission. We just do it because it's
00:20:15
important to us. And I think by default,
00:20:17
that's the kind of relationship that's
00:20:19
going to be between us and AI, which are
00:20:22
truly autonomous and operating on their
00:20:25
own behalf. And that was in 2019, the
00:20:27
year that you interviewed him.
00:20:29
>> One of the things that I I feel like we
00:20:30
should take a step back to examine is
00:20:32
going back to this idea of what even is
00:20:34
artificial intelligence and what do we
00:20:36
mean by intelligence? And a huge part of
00:20:40
the views of the different people and
00:20:41
the quotes that you're reading derives
00:20:43
from a specific belief that they each
00:20:46
have in this question of what is
00:20:49
intelligence, what constitutes
00:20:50
intelligence.
00:20:52
For Ilia, he has throughout his research
00:20:55
career felt that ultimately our brains
00:21:00
are giant statistical models. This is
00:21:03
not something that you know we actually
00:21:05
know but this is his own hypothesis also
00:21:08
the hypothesis of his mentor Jeffrey
00:21:10
Hinton who also was on this podcast.
00:21:13
This is why they have such a strong
00:21:15
conviction in the idea of building AI
00:21:18
systems that are statistical models and
00:21:20
that this particular approach is going
00:21:22
to lead to intelligent systems as we are
00:21:26
intelligent. It's a hypothesis that they
00:21:28
have. It's not one that has been proven
00:21:30
by science. And some people vehemently
00:21:33
disagree with them on this particular
00:21:35
thing. But if you step into their shoes
00:21:38
and take on that hypothesis and assume
00:21:41
that it's true, that our brains are in
00:21:44
fact statistical engines and that these
00:21:48
systems that they're building are also
00:21:50
statistical engines, that they're making
00:21:51
bigger and bigger and bigger until they
00:21:53
become the size of the human brain.
00:21:55
That's why they say that making this
00:21:59
comparison where the system will become
00:22:02
equal to human intelligence and then
00:22:03
maybe exceed human intelligence is
00:22:06
relevant in their framework. And um Ilia
00:22:09
gave a talk at one point at this really
00:22:12
prominent AI research conference that
00:22:14
happens every year called neural
00:22:16
information processing systems. It's a
00:22:18
mouthful, but he gave this keynote where
00:22:21
he shows this chart of the size of
00:22:25
brains and the intelligence of a
00:22:27
species. And it's roughly linear. The
00:22:31
bigger the size of the brain, the more
00:22:32
intelligent the species. And so for him,
00:22:36
he thinks he's building a digital brain
00:22:39
because he he thinks brains are just
00:22:41
statistical engines. So from that logic
00:22:44
it's like okay if we then build a bigger
00:22:47
statistical engine than the human brain
00:22:50
then based on this chart it will be more
00:22:53
intelligent and then we will be
00:22:55
subjected to the same treatment that
00:22:56
we've subjected animals but it's really
00:23:00
important to understand that these are
00:23:01
scientific hypotheses of specific
00:23:03
individuals within the AI research
00:23:05
community and there's a lot a lot of
00:23:08
debate about whether this is in fact the
00:23:10
case and some of The biggest critics say
00:23:14
it's very reductive to think of our
00:23:15
brains as simply just statistical
00:23:17
engines.
00:23:18
>> Why why does it matter to know the
00:23:21
mechanism?
00:23:23
Is it not just important to know the
00:23:25
outcome which is that it's going to be
00:23:27
able to do make a video for me or agents
00:23:30
are going to be able to do the work that
00:23:31
I do. Does it does it really really
00:23:33
matter for us to know the mechanism
00:23:35
behind it?
00:23:36
>> Yes and no. So it matters because these
00:23:40
companies
00:23:41
they are driving their future actions
00:23:44
based on this hypothesis.
00:23:47
So they have decided we think that this
00:23:52
hypothesis is true like we should just
00:23:54
continue building larger and larger
00:23:55
statistical models in the pursuit of
00:23:57
artificial general intelligence. And
00:24:00
that's then having global consequences
00:24:02
like in order to continue doing that
00:24:04
they're hoovering up more and more data.
00:24:07
They're building more and more data
00:24:08
centers. They are having uh they're, you
00:24:11
know, exploiting more and more labor in
00:24:13
order to continue on this path. Here's a
00:24:16
question that I think is important to
00:24:18
ask is why are we trying to build AI
00:24:21
systems that are duplicative of humans?
00:24:23
We're kind of having this conversation
00:24:24
right now where we've just taken the
00:24:27
premise of this industry as a good
00:24:30
thing. Like they said that we should be
00:24:32
building AGI, so we say that we should
00:24:34
be building AGI. I would like to ask
00:24:36
like why are we doing that? Why is it
00:24:39
that we are building a technology that
00:24:42
is ultimately designed to replace and
00:24:44
automate people away? That is not the
00:24:47
enterprise of technology. Like we should
00:24:51
be building technology and the purpose
00:24:53
of technology throughout history has
00:24:55
been to improve human flourishing, not
00:24:58
to replace people. And so this is like a
00:25:03
a critical part of my critique of these
00:25:05
companies and and these scientists that
00:25:07
have just adopted this goal and have
00:25:10
relentlessly pursued it and have had
00:25:12
enormous capital and enormous resources
00:25:13
to pursue it. Is is this the right goal?
00:25:16
What like why are we doing this? Why
00:25:18
can't we just build AI systems that do
00:25:22
things like accelerate drug discovery
00:25:24
and improve people's health care
00:25:26
outcomes, which are systems that have
00:25:28
nothing to do with the statistical
00:25:30
engines that they're trying to build to
00:25:32
duplicate the human brain?
00:25:33
>> So why are they doing it? I mean, you've
00:25:35
interviewed all these people. I think
00:25:36
it's what, 300 people in total, 80 or 90
00:25:39
of them from OpenAI, the maker of
00:25:41
CHACHBC. Why do you think they're doing
00:25:43
it?
00:25:44
I think it's because they're driven by
00:25:46
an imperial agenda. And that is why I
00:25:48
call these companies empires of AI.
00:25:50
>> What do you mean by an imperial agenda?
00:25:52
What does that term mean?
00:25:53
>> Empire is the only metaphor that I've
00:25:57
ever found to fully encapsulate all of
00:25:59
the dimensions of what these companies
00:26:01
do and the scale that they operate and
00:26:05
what motivates them to do what they do.
00:26:07
And there are many parallels that you
00:26:10
see between what I call the empires of
00:26:12
AI and the empires of old. They lay
00:26:15
claim to resources that are not their
00:26:16
own in the pursuit of training these
00:26:17
models. That's the data of individuals,
00:26:20
the intellectual property of artists,
00:26:21
writers, and creators. Their land
00:26:23
grabbing in order to build these
00:26:25
supercomputer facilities for training
00:26:27
the next generation models. Second, they
00:26:29
exploit an extraordinary amount of
00:26:30
labor. They contract hundreds of
00:26:33
thousands of workers all around the
00:26:35
world including in the US to ultimately
00:26:38
make these technologies. We can talk
00:26:40
about that more. And they also design
00:26:44
their tools to be labor automating so
00:26:46
that when the technologies are deployed,
00:26:48
it also affects labor rights because it
00:26:52
erodess away labor rights. And this is a
00:26:54
political choice that they have. Third,
00:26:57
they monopolize knowledge production.
00:26:59
And so they project this idea that
00:27:00
they're the only ones that really
00:27:01
understand how the technology works. And
00:27:03
so if the public doesn't like it, it's
00:27:05
because they don't actually know enough
00:27:06
about this technology. They do this to
00:27:08
the public. They do this to policy
00:27:10
makers. And they've also captured the
00:27:13
majority of the scientists that are
00:27:14
working on understanding the limitations
00:27:16
and capabilities of AI.
00:27:18
>> You think they're gaslighting the public
00:27:20
in a way?
00:27:20
>> They are. Yeah. So if most of the
00:27:23
climate scientists in the world were
00:27:25
bankrolled by fossil fuel companies, do
00:27:28
you think we would get an accurate
00:27:29
picture of the climate crisis?
00:27:31
>> No.
00:27:32
>> And in the same way they employ and
00:27:35
bankroll the AI industry employs and
00:27:37
bankrolls most of the AI researchers in
00:27:39
the world. So they set the agenda on AI
00:27:42
research in soft ways simply by
00:27:44
funneling money to their priorities so
00:27:47
that only certain types of AI research
00:27:49
are produced. But they also will censor
00:27:52
researchers when they do not like what
00:27:55
the researcher has found. And so I talk
00:27:58
about the case of Dr. Timmy Gabru in my
00:28:00
book who was the ethical AI team co-lead
00:28:04
at Google when she was literally hired
00:28:07
to critique the types of AI systems that
00:28:10
Google was building. She then co-wrote a
00:28:13
critical research paper that was showing
00:28:15
how large language models specifically
00:28:18
were leading to certain types of harmful
00:28:20
outcomes. And in an attempt to try and
00:28:24
stop this research from being published,
00:28:26
Google ended up firing Gabru and then
00:28:29
fired her other co-lead Margaret
00:28:31
Mitchell.
00:28:33
And so they control and quash the
00:28:38
research that is inconvenient to the
00:28:41
empire's agenda.
00:28:42
>> Did you have an example where this is
00:28:44
happening to journalists as well that
00:28:46
are asking questions of their team
00:28:48
members? I think I was watching a video
00:28:50
of yours where there was a young man
00:28:52
that was saying he had someone show up
00:28:53
at his door, knocked on his door and
00:28:55
asked for information, emails, text
00:28:58
messages, and this person was from one
00:28:59
of the big AI companies.
00:29:01
>> This was opening. I started subpoenaing
00:29:03
some of its critics. Yeah. Um as a as
00:29:06
part of a
00:29:09
what's what appears to be a campaign of
00:29:11
intimidation, but also what appeared to
00:29:12
be a campaign of fishing for more
00:29:14
information to figure out to map out the
00:29:18
network of critics further. But this was
00:29:20
a man who runs a small watchdog
00:29:24
nonprofit and they had been doing a lot
00:29:26
of work during that time to try and ask
00:29:30
questions about OpenAI's attempt to
00:29:32
convert from a nonprofit to a
00:29:34
for-profit. Ultimately, OpenAI was
00:29:36
successful in that conversion. But
00:29:37
during the period where it was sort of
00:29:40
existential for open AI to complete this
00:29:43
conversion, there were a lot of civil
00:29:45
society groups and watchdog groups like
00:29:47
MIDAS who were trying to prevent the
00:29:52
process from happening in the dead of
00:29:54
night. They were trying to get more
00:29:56
transparency. They were trying to have
00:29:57
more public debate about this because
00:29:59
it's unprecedented. And it was then that
00:30:03
um there was a knock on his door and he
00:30:06
was served papers.
00:30:08
>> What did the papers say?
00:30:09
>> The papers asked him to reproduce every
00:30:12
single piece of communication that he
00:30:15
had had that might have involved Musk.
00:30:17
So this was like this strange paranoia
00:30:18
that OpenAI had that Musk was somehow
00:30:21
funding these people to block the
00:30:23
conversion. None of them were actually
00:30:24
funded by Musk. So in this particular
00:30:27
case their request he simply was just
00:30:29
answered you know I I don't have any
00:30:31
documents because this doesn't exist.
00:30:33
>> So going back to this point of empires
00:30:35
you were saying that one of the factors
00:30:36
of an empire is a land grab and then the
00:30:39
next one was
00:30:40
>> was labor exploitation
00:30:42
>> labor exploitation. The third one,
00:30:44
controlling knowledge production.
00:30:47
>> And one of the other ones that's really
00:30:50
important to understand about the AI
00:30:52
empires in particular is empires always
00:30:56
have this narrative that they they say
00:30:59
to the public like we're the good empire
00:31:02
and we need to be an empire in the first
00:31:04
place because there are also bad empires
00:31:06
in the world. And if you allow us to
00:31:10
take all the resources and use all of
00:31:12
the labor, then we promise we will bring
00:31:15
you progress and modernity for everyone.
00:31:18
>> We will bring you to this utopic state
00:31:20
akin to an AI heaven. But if the evil
00:31:24
empire does it first, we will descend
00:31:26
into a hell.
00:31:28
>> And the evil empire being in this case,
00:31:30
>> in this case, most often it's China. But
00:31:33
actually in the early days, Open AI
00:31:35
evoked Google as the evil empire.
00:31:38
>> So all of their decisions were about we
00:31:40
need to do it first because otherwise
00:31:42
Google, this evil corporation that's
00:31:44
driven by profit, us as a benevolent
00:31:47
nonprofit. Like this is a this is a
00:31:50
critical contest of who wins.
00:31:54
>> Do you think the people building these
00:31:56
AI companies believe that the outcome is
00:32:00
going to be all good now? Do you think
00:32:02
they think that it's going to be it's
00:32:04
going to serve everyone? It's going to
00:32:05
be the age of abundance. Everything's
00:32:07
going to go up well. What do you think
00:32:08
they believe? What do you think Sam
00:32:09
believes?
00:32:10
>> So, so this is so funny is such a core
00:32:13
part of the mythology that they create
00:32:15
around the AI industry includes the
00:32:19
belief that it could go very badly. It
00:32:22
goes hand in hand. like they need that
00:32:25
part of the myth in order to then say
00:32:28
and that's why we need to be in control
00:32:30
of the technology because that's the
00:32:32
only way that it's going to go really
00:32:33
really well and Alman has said publicly
00:32:35
you know the worst case lights out for
00:32:38
everyone but best case we cure cancer we
00:32:42
solve climate change and there's
00:32:43
abundance and Dario Amade same kind of
00:32:46
rhetoric was like worst case
00:32:49
catastrophic or existential harm for
00:32:52
humanity best case mass human
00:32:55
flourishing. So this is like two sides
00:32:57
of the same coin. Like they have to use
00:33:00
both of these narratives in order to
00:33:03
continue justifying an extremely
00:33:06
anti-democratic approach to AI
00:33:07
development where there should not be
00:33:10
broad participation in developing this
00:33:12
technology. They must be the ones
00:33:14
controlling it at every step of the way.
00:33:16
>> Sam Orman did a tweet saying, "There are
00:33:18
some books coming out about open AI and
00:33:20
me. We only participated in two of them.
00:33:23
one by Kesh Hegy
00:33:25
>> Keegy
00:33:26
>> Khaggy focused on me and one by Ashley
00:33:29
Vance on OpenAI.
00:33:31
Um he went on to say no book will get
00:33:33
everything right especially when some
00:33:35
people are so intent on twisting things
00:33:38
but these two authors are trying to
00:33:41
you quote retweeted that tweet from Sam
00:33:44
Alman and you said the unnamed book
00:33:47
empire of AI is mine.
00:33:51
Do you believe that tweet from Sam Alman
00:33:53
was in reference to your book?
00:33:55
>> 100%. Because there's only three books
00:33:57
coming out about him
00:33:58
>> and he had caught wind that your book
00:33:59
was coming out and
00:34:00
>> he knew my book was coming out because I
00:34:02
had contacted OpenAI from the very
00:34:04
beginning of my process and said I'm
00:34:05
working on a book now. Will you
00:34:07
participate in it? And actually
00:34:09
initially they said yes even though so
00:34:12
my history with OpenAI I profiled the
00:34:14
company for MIT technology review. I
00:34:16
embedded within the office for 3 days in
00:34:19
2019. my profile comes out in 2020, the
00:34:22
leadership are very unhappy. And in my
00:34:25
book, I actually quote an email that I
00:34:27
received that Sam Alman sent to the
00:34:30
company about my profile saying, "Yeah,
00:34:33
this is not great."
00:34:36
And from then on, the company's stance
00:34:40
to me was,
00:34:43
"We are not going to participate in
00:34:46
anything that you do. we are not going
00:34:48
to respond to anything any of the
00:34:49
questions that you receive. And this
00:34:51
was, you know, this was things that they
00:34:54
explicitly articulated. It wasn't like
00:34:56
me inferring. Um, so I I had a a
00:34:59
colleague at MIT Technology Review that
00:35:01
also covered AI. And at one point
00:35:03
opening, I sent him this press release
00:35:05
being like, "We would love for you to
00:35:06
cover this story." And he was like, "I'm
00:35:07
really busy. Will you send it to Karen?"
00:35:10
And they were like, "Oh, no. We have a
00:35:13
history. You understand?" And so, so for
00:35:17
three years they they refused to talk to
00:35:19
me, but then I ended up at the Wall
00:35:22
Street Journal where if they felt a a
00:35:25
bit compelled because it was the journal
00:35:27
to reopen the lines of communication.
00:35:30
And so I I I started having, you know,
00:35:33
more dialogue with them. Every time I
00:35:35
wrote a piece, I would always send them
00:35:37
here's my request for comment. I would
00:35:39
always ask them like, will you sit for
00:35:40
interviews? And we did get to a more
00:35:43
productive relationship. And then I
00:35:45
embarked on the book. So I I left the
00:35:47
journal to focus on the book full-time.
00:35:49
And I told them right away, I'm working
00:35:52
on this book. I want to continue this
00:35:55
productive conversation where I make
00:35:58
sure I reflect OpenAI's perspective in
00:36:01
the book. And so they were like, we can
00:36:04
arrange interviews for you. You can come
00:36:05
back to the office. We'll set up some
00:36:09
conversations.
00:36:10
And then as we were going back and forth
00:36:13
on this, the board fired Sam Alman.
00:36:17
And that's when things started going
00:36:19
kind of south because the company
00:36:21
started becoming very sensitive to
00:36:23
scrutiny. And so then they started
00:36:25
pushing kicking the can down the road,
00:36:27
down the road, down the road. And I kept
00:36:28
saying, "Hey, when are we rescheduling
00:36:30
this? What's going on?" And then I get
00:36:32
an email saying, "We are not going to
00:36:34
participate at all. You are not coming
00:36:36
to the office. You're not doing
00:36:37
interviews." and I had actually already
00:36:39
booked my tickets. So, I was already
00:36:41
going to fly to San Francisco to have
00:36:44
the the interviews. And so, then I told
00:36:48
them I was like, "That's fine. I will
00:36:51
still engage in the process where I'll
00:36:53
give you extensive requests for comment.
00:36:55
I'll ask through my reporting, I'll keep
00:36:57
you updated on all the things that I'm
00:36:59
finding so that you can choose to still
00:37:01
comment." I gave them 40 pages of
00:37:04
requests for comment. and I gave them
00:37:07
over a month to respond to all of that.
00:37:09
So, this was when the tweet came out was
00:37:12
we were doing all this back and forth
00:37:14
trying to
00:37:15
and that's when Alman tweeted this.
00:37:20
>> H
00:37:21
>> and they never responded to a single one
00:37:23
of the one of the 40 pages.
00:37:25
>> Sam Alman does a lot of interviews.
00:37:27
>> Yeah.
00:37:28
>> You know, he's doing a lot of interviews
00:37:29
all the time. He's done every podcast.
00:37:31
I've seen him on everything from Tucker
00:37:33
Carlson to I think he's done Theo, Joe
00:37:35
Rogan, um podcasts all over the world.
00:37:39
>> I wonder why he won't do mine.
00:37:45
>> Well, maybe.
00:37:46
>> I don't know why. I I I don't know. I
00:37:48
think I'm fair with everyone. I just ask
00:37:49
I just ask questions I genuinely care
00:37:50
about. I don't come in with huge
00:37:52
preconceptions or at least meet people
00:37:54
for the first time. But I've heard
00:37:56
through the grape vine
00:37:58
um that he doesn't want to do mine. I
00:38:00
mean, going back to what you were saying
00:38:02
earlier that
00:38:04
with this the way that OpenAI and these
00:38:06
companies control research, you asked,
00:38:09
do they also do this with journalists?
00:38:12
I mean, yes, the answer is yes. And
00:38:14
apparently they they also do it with
00:38:16
anyone who has, you know, a broad mass
00:38:18
communications platform.
00:38:20
>> It's not just about the conversation
00:38:22
that you're going to have with them.
00:38:24
It's about who you also choose to
00:38:26
platform.
00:38:28
And there's this huge problem in
00:38:30
technology journalism where companies
00:38:33
know that a really big carrot that they
00:38:37
can give to technology journalists is
00:38:38
access.
00:38:39
>> Yeah. Yeah. Yeah.
00:38:40
>> And they will withhold that access at
00:38:44
the drop of a hat if they catch wind
00:38:46
that you're speaking to someone that
00:38:47
they didn't want you to speak to.
00:38:49
>> This is so true. And I don't think the
00:38:51
average person really truly understands
00:38:53
this.
00:38:54
>> Yeah. So, this kind of sounds like
00:38:55
theory as you say it, but I'm not going
00:38:57
to name names here because I don't think
00:38:59
it's important, but there is a
00:39:01
particular person in AI who um whose
00:39:05
team have basically dangled the carrot
00:39:08
of them coming here for like 18 months.
00:39:10
And I'm like, you don't you don't have
00:39:11
to dangle the carrot. I'm going to speak
00:39:13
to whoever I want to regardless of the
00:39:14
carrot or not. And when this person
00:39:16
comes, if they want to come, I'll I'll
00:39:17
give them a fair shot. I'll ask them all
00:39:19
genuinely curious questions about what
00:39:21
they're doing, their incentives. I won't
00:39:23
gotcha them. I don't have a history of
00:39:24
ever gotchering anybody. Even if I dis
00:39:26
like even if I have a different of
00:39:28
opinion, I'll ask the question.
00:39:29
>> Yeah.
00:39:29
>> But they dangle carrots and they say,
00:39:31
"Well, if you know he he's thinking
00:39:33
about it, let's think about a date." And
00:39:34
what what the strategy is, and I don't
00:39:36
think they they think those people don't
00:39:37
understand, is if we just dangle it for
00:39:39
long enough, then they will
00:39:42
um perform in the way that we want them
00:39:44
to do and they'll be
00:39:46
>> they'll be pleasant about us. They won't
00:39:49
be critical. They won't give a give a
00:39:52
critics.
00:39:52
>> Our critics.
00:39:53
>> And I think a lot of their game is just
00:39:55
dangle the carrot forever.
00:39:57
>> Yes. Yeah.
00:39:57
>> That's like the optimal outcome is if we
00:39:59
just dangle it. If we just tell them,
00:40:00
yeah, look, we're just trying looking at
00:40:01
the schedule.
00:40:03
>> It just doesn't work. I think in the
00:40:04
modern world, you just have to go there
00:40:05
and give your opinion and allow the
00:40:07
clash of ideas in the public forum, let
00:40:08
the viewers un decide for themselves.
00:40:10
>> Yeah.
00:40:11
>> What they think.
00:40:12
>> Yeah.
00:40:12
>> Um, but this is a Yeah. This is such a
00:40:14
huge part of their machinery is the way
00:40:18
that they use these tactics to massage
00:40:21
the public image of these companies and
00:40:23
make sure that information that they
00:40:24
don't want out and even opinions that
00:40:26
they don't want out there go out there.
00:40:28
>> Mhm.
00:40:29
>> And so this is this is you know I feel
00:40:33
very lucky now that opening I shut the
00:40:36
door early on me
00:40:38
>> at the time I didn't feel lucky. I felt
00:40:40
like I had screwed myself over. I was
00:40:43
nicer
00:40:45
access
00:40:47
to a journalist, right? Like you're
00:40:49
supposed to report the truth and you're
00:40:51
always supposed to report in the
00:40:53
interest of the public. Like that is the
00:40:55
point of journalism. And in that moment
00:40:58
it I I was like relatively junior in my
00:41:00
career. I was like, did I misunderstand
00:41:03
what journalism about is is about? Like
00:41:06
>> should I have actually been playing the
00:41:08
access game?
00:41:09
>> Mhm.
00:41:09
>> But it was too late. I had the door shut
00:41:11
to me and so I had to build my career
00:41:15
understanding that the door the front
00:41:16
door was never going to be open.
00:41:18
>> Yeah.
00:41:19
>> And that actually really strengthened my
00:41:22
own ability to just tell it like it is
00:41:26
like objective. Yeah. And just report
00:41:28
what I see are the facts being presented
00:41:31
to me irrespective of whether the
00:41:33
company likes it or not. And most often
00:41:35
the company really does not like it but
00:41:38
>> I can continue to do the work. They
00:41:40
don't need to open the front door for
00:41:41
me. I was still able to do more than 300
00:41:44
interviews.
00:41:45
>> So Sam Alman gets
00:41:49
kicked off the OpenAI executive team.
00:41:55
Did you find out why that happened?
00:41:57
>> Yeah, there's a
00:42:00
scene by scene recounting
00:42:02
>> from who? I can't remember the exact
00:42:04
number of sources, so I don't want to
00:42:06
misquote myself, but it was around six
00:42:08
or seven people that were directly
00:42:10
involved or had spoken to people
00:42:11
directly involved in the decision-making
00:42:13
process.
00:42:15
So,
00:42:18
Ilia Satskever
00:42:20
is seeing these serious concerns about
00:42:24
the way that Altman's behavior is
00:42:27
leading to
00:42:29
bad research outcomes and poor
00:42:32
decision-m at the company.
00:42:35
He then approaches a board member, Helen
00:42:38
Toner. Ilia, for anyone that doesn't
00:42:40
know, is the the co-founder we mentioned
00:42:42
earlier. The co-founder of OpenAI we
00:42:44
mentioned earlier.
00:42:45
>> Yes. And he kind of does a bit of a
00:42:50
sounding board thing to Helen just
00:42:51
because Ilia is freaking out. He's like
00:42:54
he's been like sitting on this these
00:42:56
these concerns for a while and he's like
00:42:58
if I tell this to someone, this could
00:43:01
also be really bad for me if Alman finds
00:43:05
out.
00:43:06
And so he asks for a meeting with Toner
00:43:12
and in that first meeting he's like
00:43:15
re like he barely says a thing. He's
00:43:18
just like dancing around trying to
00:43:20
figure out hey is this someone that I
00:43:23
can maybe trust to divulge more
00:43:25
information.
00:43:25
>> And Toner's role and responsibilities at
00:43:27
OpenAI were
00:43:28
>> she was a board member.
00:43:29
>> Just a board member.
00:43:30
>> Yeah. And and specifically an
00:43:31
independent board member. So opening eye
00:43:34
when it was a nonprofit the board was
00:43:36
split between people who had a stake
00:43:38
financial stake in the company and then
00:43:40
people who were fully independent and
00:43:42
this was meant to be a structure that
00:43:43
would balance the decision-m to be in
00:43:47
the benefit of the public interest
00:43:48
rather than to be in the benefit of the
00:43:49
for-profit entity that opening I then
00:43:51
created
00:43:52
>> and
00:43:54
Ilia as a
00:43:57
non-independent board member was
00:43:59
approaching toner as an independent
00:44:01
board member her to try and see whether
00:44:05
or not she was potentially seeing or
00:44:08
hearing the same things that he was
00:44:10
about the effect that Alman was having
00:44:12
on the company. This then sets off a
00:44:14
series of conversations first between
00:44:17
Ilia and Helen and then between Amir
00:44:21
Moratti and some of the board members.
00:44:23
Samir Moratti was at that point the
00:44:25
chief technology officer of OpenAI where
00:44:28
these two senior leaders essentially
00:44:30
through these conversations and through
00:44:31
documentation that they're pulling
00:44:33
together like email, Slack messages and
00:44:35
so forth, they convey to the independent
00:44:37
board members, three independent board
00:44:39
members, we are very concerned about
00:44:44
Altman's leadership like he is creating
00:44:47
too much instability at the company and
00:44:51
it is like he is the root of the
00:44:54
problem. It's not they they they were
00:44:57
trying to say to these independent board
00:44:58
members like the problem will not be
00:45:01
fixed unless Alman is removed because of
00:45:04
the way that he's pitting teams against
00:45:06
each other and creating this environment
00:45:09
where people are unable to trust each
00:45:10
other anymore and they're competing
00:45:12
rather than collaborating on what's
00:45:14
supposed to be this really really
00:45:15
important technology. When you say
00:45:18
instability,
00:45:20
that's a that's quite a vague term. That
00:45:21
could mean lots of things. Like
00:45:23
instability could mean pushing people
00:45:24
hard to work harder,
00:45:25
>> right?
00:45:26
>> What do you mean by instability in spec
00:45:28
as specific terms as you can possibly
00:45:30
say them?
00:45:31
>> When chat GBT came out in the world,
00:45:34
OpenAI was wholly unprepared.
00:45:36
>> They didn't think that they were
00:45:38
launching a gangbusters product.
00:45:41
>> Yeah. They thought they were releasing a
00:45:43
research preview that would help them
00:45:46
get the data flywheel going, collect a
00:45:48
bunch of data from users that would then
00:45:50
inform what they thought would be the
00:45:53
gang busters product, which was a
00:45:55
chatbot using GPT4 and chat GBT was
00:45:59
using GPT 3.5.
00:46:01
And because of that, there were servers
00:46:06
crashing all the time because they they
00:46:08
weren't they had to scale their their
00:46:10
infrastructure, you know, faster than
00:46:12
any company in history. And there were
00:46:15
um there were all of these outages. They
00:46:17
were trying to also hire faster than any
00:46:19
company in history to try and have more
00:46:20
personnel there. And they were then
00:46:23
sometimes hiring people that they were
00:46:24
like, "Actually, we made a mistake. We
00:46:26
shouldn't have hired you." So they were
00:46:27
firing people left and right. and people
00:46:29
were just disappearing off of Slack and
00:46:32
that's how their colleagues would learn
00:46:33
that they were no longer at the company.
00:46:35
And so it was yes like many fast growing
00:46:39
companies a very chaotic environment and
00:46:42
a particularly chaotic environment
00:46:44
because it was extra fast like they had
00:46:48
to accelerate more than any other
00:46:51
startup.
00:46:52
And on top of that mirror Morati and
00:46:55
Ilasgiver felt that Alman was making it
00:46:58
worse like he was not actually
00:47:00
effectively ameliorating the
00:47:02
circumstances of the chaos. He was
00:47:05
actually sewing more chaos, getting
00:47:06
these teams to be more divided.
00:47:10
And this is where it's important to
00:47:13
understand that the executives and the
00:47:16
independent board members, they're all
00:47:19
operating under this idea that they're
00:47:21
building AGI and that AGI could either
00:47:24
be devastating or utopic to humanity.
00:47:29
And so it's not yes it's like any other
00:47:32
company and no it's not like any other
00:47:34
company. You cannot have like in their
00:47:37
view you cannot have this degree of
00:47:39
chaos as the pressure cooker for
00:47:42
creating a technology that they in their
00:47:44
conception could make or break the
00:47:47
world.
00:47:48
And so that is basically what the
00:47:51
independent board members also begin to
00:47:53
reflect on. They have these
00:47:54
conversations amongst themselves where
00:47:56
they're like,
00:47:58
"Well, based on what we're hearing about
00:48:00
Altman's behavior, like if this was an
00:48:02
Instacart, would that warrant firing
00:48:04
him?" And they concluded, "Maybe not,
00:48:08
but this is not Instacart."
00:48:10
And that's why they were like, "Well,
00:48:12
crap. Maybe this is actually this does
00:48:15
rise to the to the bar where we should
00:48:18
consider replacing him because we are
00:48:21
ultimately building a technology that we
00:48:24
think could have transformative impacts
00:48:27
either in the positive or negative
00:48:29
direction. And so that is what happens.
00:48:31
It's like these two executives and then
00:48:33
the independent board members also they
00:48:35
were hearing other feedback as well from
00:48:37
their connections within the company
00:48:38
with other people in the industry. At
00:48:40
one point, Adam D'Angelo, who is one of
00:48:42
the independent board members and the
00:48:44
CEO of Kora, uh, which is, you know,
00:48:46
start a tech startup in the valley, he
00:48:49
is at a party in San Francisco, and he
00:48:52
starts to hear some of these rumors that
00:48:56
there's something weird about the way
00:48:58
that OpenAI has structured its OpenAI
00:49:02
startup fund, which was this fund that
00:49:04
they the company had created to start
00:49:06
investing in other startups.
00:49:08
>> Mhm.
00:49:09
and he realizes they'd never really seen
00:49:13
documentation about how the startup fund
00:49:15
had been set up from Alman. And finally
00:49:17
they get the documents and it turns out
00:49:18
that OpenAI startup fund is not OpenAI's
00:49:21
startup fund. It's Altman's startup
00:49:23
fund. And this was something like one of
00:49:28
several experiences that the independent
00:49:29
board members were also having where
00:49:31
they're like there's something not right
00:49:33
about the fact that there continuously
00:49:37
are inconsistencies inconsistencies
00:49:39
between the way that Altman is
00:49:40
portraying
00:49:42
what is being done versus what is
00:49:44
actually being done. And so when these
00:49:47
two executives approach the board or the
00:49:49
independent board members, then they're
00:49:51
like, "Okay, this lines up with also the
00:49:54
experiences that we've been having."
00:49:58
And at that point, they then have this
00:50:01
series of very intense discussions where
00:50:04
they're meeting almost every day talking
00:50:06
about should we actually really consider
00:50:09
removing Altman?
00:50:12
And in the end they conclude, yes, we
00:50:16
should. And if we're going to do it, we
00:50:18
need to do it quickly. Because they were
00:50:20
very concerned that the moment that
00:50:22
Alman found out, his persuasive
00:50:23
abilities would make it impossible to
00:50:26
do. And so they end up firing Altman
00:50:31
without telling anyone. You know, they
00:50:33
don't talk to any stakeholders to get
00:50:36
them on the same page. Microsoft gets a
00:50:38
call right before they execute the
00:50:40
action saying, "We're going to fire
00:50:42
Altman."
00:50:42
>> And Microsoft, for anyone that doesn't
00:50:43
know, are a lead investor in OpenAI at
00:50:45
the time.
00:50:46
>> Yes. One of the only investors in OpenAI
00:50:51
at the time. And that is what then
00:50:54
devolves the whole thing because every
00:50:57
single person that is affected by this
00:50:59
decision is now extremely angry that
00:51:02
they were not involved. And that is what
00:51:05
then creates this campaign to bring
00:51:08
Altman back. And then Alman is
00:51:10
reinstalled as CEO days later.
00:51:13
>> This company that I've just invested in,
00:51:15
it's grown like crazy. I want to be the
00:51:16
one to tell you about it because I think
00:51:17
it's going to create such a huge
00:51:18
productivity advantage for you. Whisper
00:51:20
Flow is an app that you can get on your
00:51:22
computer and on your phone on all your
00:51:23
devices and it allows you to speak to
00:51:25
your technology. So, instead of me
00:51:26
writing out an email, I click one button
00:51:28
on my phone and I can just speak the
00:51:30
email into existence and it uses AI to
00:51:33
clean up what I was saying. And then
00:51:34
when I'm done, I just hit this one
00:51:36
button here and the whole email is
00:51:37
written for me. And it's saving me so
00:51:40
much time in a day because Whisper
00:51:42
learns how I write. So on WhatsApp, it
00:51:44
knows how I am a little bit more casual.
00:51:46
On email, a little bit more
00:51:47
professional. And also, there's this
00:51:48
really interesting thing they've just
00:51:49
done. I can create little phrases to
00:51:51
automatically do the work for me. I can
00:51:53
just say Jack's LinkedIn and it copies
00:51:55
Jack's LinkedIn profile for me because
00:51:57
it knows who Jack is in my life. This is
00:51:58
saving me a huge amount of time. This
00:52:00
company is growing like absolute crazy.
00:52:02
And this is why I invested in the
00:52:03
business and why they're now a sponsor
00:52:04
of this show. And Whisper Flow is
00:52:06
frankly becoming the worstkept secret in
00:52:08
business, productivity, and
00:52:09
entrepreneurship. Check it out now at
00:52:11
whisperflow spelled w i s p r l o
00:52:15
w.ai/steven.
00:52:17
It will be a game changer for you.
00:52:20
There's a phase a lot of companies hit
00:52:21
where they're no longer doing the most
00:52:23
important thing, which is selling. And
00:52:24
they get really bogged down with admin.
00:52:26
And it's often something that creeps up
00:52:28
slowly and you don't really notice until
00:52:30
it's happened. Slowly momentum starts to
00:52:32
leak out. This happened to us and our
00:52:34
sponsor Pipe Drive was a fix I came
00:52:36
across 10 years ago. And ever since my
00:52:38
teams across my different companies have
00:52:39
continued to use it. Pipe Drive is a
00:52:41
simple but powerful sales CRM that gives
00:52:43
you the visibility on any deals in your
00:52:45
pipeline. It also automates a lot of the
00:52:47
tedious, repetitive, and time-conuming
00:52:49
parts of the sales process, which in
00:52:51
turn saves you so many hours every
00:52:52
single month, which means you can get
00:52:54
back to selling. Making that early
00:52:55
decision to switch to Pipe Drive was a
00:52:57
real gamecher, and it's kept the right
00:52:59
things front of mind. My favorite
00:53:01
feature is Pipe Drive's ability to sync
00:53:03
your CRM with multiple email inboxes so
00:53:05
your entire team can work together from
00:53:07
one platform. And we aren't the only
00:53:09
ones benefiting. Over 100,000 companies
00:53:11
use Pipe Drive to grow their business.
00:53:13
So, if something I've said resonates,
00:53:15
head over to pipedive.com/ceeo
00:53:18
where you can get a 30-day free trial.
00:53:21
No credit card or payment required.
00:53:24
How does a CEO of a major company get
00:53:27
fired by the board? Because board
00:53:29
members, there's a quote in your book on
00:53:30
page 357 where you say about Ilia
00:53:33
saying, "I don't think Sam is the guy
00:53:35
who should have the finger on the button
00:53:36
for AGI." Now, I I asked myself this
00:53:39
question. You know, I work with lots of
00:53:40
people here. We have 150 people that
00:53:42
work in this business and
00:53:46
those people know me best.
00:53:48
>> Yeah.
00:53:48
>> They see me on camera. They see me off
00:53:50
camera. So if they said that we don't
00:53:52
think Steven is the right person to host
00:53:54
the direc
00:53:55
>> Yeah.
00:53:56
>> It would take a lot for them to say
00:53:58
that.
00:53:58
>> Yeah.
00:53:58
>> They must have seen some [ __ ] off camera
00:54:01
for them to go we don't think he's the
00:54:02
right person to be on camera. Yeah.
00:54:04
>> Or for whatever reason. And in the case
00:54:05
of AI, which is much more consequential
00:54:07
than a podcast that is, you know, filmed
00:54:09
in my old kitchen. Um it almost sends a
00:54:12
chill down one's body to think that the
00:54:14
co-founder of a business has gone to the
00:54:16
board and said this isn't the guy to
00:54:18
lead this consequ I mirror Marotti then
00:54:21
also said I don't think Alman is the
00:54:23
right guy
00:54:23
>> and then they both left later.
00:54:26
>> So then Altman comes back and lo and
00:54:28
behold Ilia never comes back. So his
00:54:31
concerns about the fact that Alman
00:54:33
founding out would be bad for him
00:54:35
manifested. He ended up not coming back
00:54:37
and Miriam Marotti then left shortly
00:54:40
thereafter.
00:54:41
>> Quite a lot of these people leave, don't
00:54:43
they? Open AAI
00:54:44
>> they do. So if you consider
00:54:49
one of the
00:54:51
origin stories of open AI is this dinner
00:54:54
that happened at the Rosewood Hotel,
00:54:57
which is a very swanky hotel um right
00:54:59
right in the heart of Silicon Valley
00:55:01
that uh was one of Elon Musk's favorites
00:55:04
whenever he was coming up from LA to the
00:55:06
Bay Area. And there was this dinner that
00:55:08
was there where Altman was intending to
00:55:11
recruit the OG team that would start
00:55:14
OpenAI. So he's kind of telling everyone
00:55:18
you might have a chance to meet Musk
00:55:20
because Musk is going to come to this
00:55:21
dinner dinner. And he cold emails Ilia
00:55:24
and gets Ilia to then come because and
00:55:27
Ilia specifically wants to come because
00:55:28
he wants to meet Musk. And he also
00:55:31
emails all these other people including
00:55:33
Greg Brockman, Dario Amade. These are
00:55:35
all people that ended up working at Open
00:55:37
>> and they all almost all of them not not
00:55:40
every one of them but almost all of them
00:55:42
end up working at OpenAI
00:55:45
>> and leaving
00:55:46
>> almost all of them end up leaving
00:55:49
specifically after they clash with Alman
00:55:52
>> and Ilia he left and launched a company
00:55:55
called Safe Super Intelligence.
00:55:59
>> Yeah.
00:56:00
>> Which is I mean that's an indirect if
00:56:02
I've ever heard one. Do you know what I
00:56:05
mean? Do you know what I mean? If
00:56:07
someone like co-ounded this podcast with
00:56:10
me and then they left and started a
00:56:12
podcast called Safe Podcasting, I
00:56:17
I'd take that as a slight.
00:56:19
I' I'd have people knocking on their
00:56:21
door and asking for their texts. One of
00:56:24
the things that is happening here is
00:56:30
>> it is not a coincidence that every
00:56:32
single tech billionaire has their own AI
00:56:34
company.
00:56:35
>> Mhm.
00:56:38
>> They want to create AI in their own
00:56:40
image and that's why they keep not
00:56:44
getting along. And in fact, it's not
00:56:46
just don't get along, they end up hating
00:56:49
each other after working together.
00:56:51
>> Mhm. and then splinter off into their
00:56:54
own organizations. So after Musk leaves,
00:56:57
he starts XAI. After Dario leaves, he
00:56:59
starts Anthropic. After Ilia leaves, he
00:57:01
starts Safe Super Intelligence. After
00:57:03
Meera leaves, she starts thinking
00:57:05
machines lab. They want to have control
00:57:12
over their own vision of this
00:57:14
technology. And the best way that they
00:57:18
have
00:57:20
derived from their experiences of trying
00:57:24
to put their vision into the arena is by
00:57:28
creating a competitor and then competing
00:57:30
with OpenAI and with all the other
00:57:32
companies out there. Do you think some
00:57:33
of these AICOs realize that they are
00:57:35
quite literally summoning the demon as
00:57:36
Elon said 10 years ago, but they don't
00:57:40
really care because being the person
00:57:42
that summoned the demon is makes you
00:57:45
consequential and powerful and
00:57:47
historical even if the outcome is
00:57:50
potentially horrific. Even if there's
00:57:52
like a 20% outcome of it being horrific.
00:57:53
I remember I think it was Dario, he's
00:57:56
the one that said there's somewhere
00:57:57
between a 10% and 25% chance of things
00:58:02
going catastrophically wrong on the
00:58:04
scale of human civilization. 25% is a
00:58:07
one in4 chance.
00:58:10
If you put bullets in a fourchamber
00:58:13
revolver and said Steven, the upside is
00:58:17
you could become a multi-gazillionaire
00:58:19
and be remembered forever. The downside
00:58:21
is that there would be a bullet in your
00:58:22
head. There is no chance that I would
00:58:24
take take that bet with a 25% potential
00:58:27
chance of things going catastrophically
00:58:30
wrong.
00:58:31
>> So, I have a very long answer to this
00:58:33
because
00:58:36
do they know if they're summoning the
00:58:37
demon? It really depends on what we
00:58:38
define as summoning the demon. And in
00:58:41
this particular case, to go back to what
00:58:44
we were saying before, there's a
00:58:46
mythology that the AI industry uses
00:58:50
where summoning the demon is an integral
00:58:52
part of
00:58:55
convincing everyone that therefore they
00:58:58
can be the only ones that are developing
00:59:00
this technology.
00:59:01
>> I got it. So on one end, you got to say
00:59:03
if we don't, China will and that's
00:59:05
terrible.
00:59:06
>> Yeah. But if we let anyone else do it
00:59:08
other than me, then we're [ __ ] as
00:59:09
well.
00:59:10
>> Exactly.
00:59:11
>> So that means that I have to do it and
00:59:12
you have to give me money and support.
00:59:14
>> Exactly. So when they're saying these
00:59:15
things,
00:59:18
we should understand it as not as like a
00:59:21
genuine prediction based on what they're
00:59:23
seeing because first of all, we don't
00:59:24
predict the future. We make it. We
00:59:27
should understand this as an act of
00:59:29
speech to persuade other people into
00:59:32
believing that they should seed more
00:59:34
power, more resources to these
00:59:36
individuals. And so, do they know that
00:59:39
they're summoning the demon?
00:59:41
I mean, they are purposely trying to
00:59:43
create this this
00:59:47
feeling within the public that they are
00:59:49
because it is a crucial part of their
00:59:52
power.
00:59:53
But do they if we were to define
00:59:57
just do they realize that the things
00:59:59
that they are doing are having already
01:00:01
really harmful impacts all around the
01:00:03
world on vulnerable people, vulnerable
01:00:06
communities, vulnerable countries.
01:00:09
That's where I'm like maybe yes, maybe
01:00:11
no. and they don't really care because
01:00:15
in the frame of mind like I sometimes
01:00:19
use the analogy that the AI world is
01:00:21
like Dune.
01:00:22
>> Dune for anyone that doesn't know Dune
01:00:24
>> science fiction epic written by Frank
01:00:25
Herbert and it's set in this
01:00:28
intergalactic era where there are all
01:00:30
these houses and they're fighting each
01:00:32
other for spice. So it's a call back to
01:00:34
colonialism and empire and they all are
01:00:37
trying to control the spice. But one of
01:00:38
the features of this story is that there
01:00:41
are these myths that are seated on the
01:00:44
different planets about a a religious
01:00:47
myth basically about the coming of the
01:00:48
Messiah that are used as ways to control
01:00:51
the people.
01:00:52
And Paul at Trades when he arrives at
01:00:56
the planet Iraqis uh with with the
01:00:59
intention of um trying to then fight
01:01:01
against the empire and um avenge his
01:01:06
father's death. He steps into a myth
01:01:09
that has been seated on this planet that
01:01:11
says that one day there will be a
01:01:13
Messiah that comes and saves the planet.
01:01:15
So he steps into the role of the Messiah
01:01:18
and leans into this idea in order to
01:01:21
better control the people and rally them
01:01:24
behind him as a leader to help with this
01:01:27
quest.
01:01:29
He knows that it's a myth in the
01:01:30
beginning, but because he lives and
01:01:33
breathes and embodies it, it kind of
01:01:37
starts to blur in his mind whether this
01:01:39
is really a myth or whether he's really
01:01:40
the messiah. And this is what I think
01:01:44
happens in the AI world. On one hand,
01:01:48
there are all these executives that
01:01:51
actively engage in mythmaking because,
01:01:54
you know, I have all these internal
01:01:55
documents that I write about in the book
01:01:57
where they are very keenly aware of how
01:02:00
to bring the public along with them by
01:02:03
showing them dazzling demonstrations of
01:02:06
the technology by using crafting a
01:02:09
mission that will sound really good uh
01:02:12
and and and make people give more
01:02:15
leniency to their companies. So they
01:02:18
know they're doing the mythmaking and
01:02:20
also I think many of them lose
01:02:23
themselves in the myth because they have
01:02:26
to live and breathe and embody it day in
01:02:28
and day out. And so when you know Daario
01:02:31
says he thinks that 10 to 25% of the
01:02:35
future could be catastrophic or or
01:02:37
whatever the probability is 10 to 25%.
01:02:40
He is actively engaging in the
01:02:41
mythmaking but also he's losing himself
01:02:44
in the myth. Like I think if you were to
01:02:46
ask him, "Do you genuinely believe
01:02:47
that?" He would be like, "Yes, I
01:02:49
genuinely believe that." Because there's
01:02:51
been a blurring of when he's saying
01:02:54
something just to say something versus
01:02:57
when he actually believes what is he's
01:03:01
required to believe in order to then
01:03:04
continue
01:03:06
doing the things that he's doing.
01:03:09
>> And this is the whole psychology of
01:03:11
cognitive dissonance, right? where you
01:03:12
the brain struggles to hold two
01:03:14
conflicting worldviews at the same time.
01:03:16
So it's it's incentivized or it
01:03:18
endeavors to dismiss one. So if you you
01:03:20
know if you wanted to be a healthy
01:03:21
person but also a smoker. Um and I
01:03:24
pointed out that smoking is bad for you.
01:03:25
The first words out of your mouth are
01:03:26
going to be yes but
01:03:28
>> smoking helps me with stress. Yeah, but
01:03:31
I only do it when I think I don't know I
01:03:34
kind of see that at the moment because
01:03:35
these companies have to raise
01:03:37
extortionate like huge amounts of money
01:03:39
to fund their AI research and they're
01:03:42
building out all of these data centers.
01:03:44
>> So when they're out in the public,
01:03:46
they're always fundraising. All of these
01:03:47
major companies are fundraising all the
01:03:48
time at the moment.
01:03:49
>> So you can't be fundraising and saying,
01:03:51
"I'm going to destroy your children's
01:03:52
future potentially. There's 25% chance
01:03:54
that your children aren't going to have
01:03:56
a great life."
01:03:58
Which might be the truth. I mean that is
01:03:59
actually what they say Dario. This is
01:04:01
what famously Dario Amade does. He's
01:04:03
like
01:04:03
>> he does that but the others Sam's not
01:04:05
doing that as much anymore.
01:04:06
>> Yes. And it's because you know
01:04:10
it goes back to like each of them kind
01:04:11
of distinguish themselves a little bit
01:04:13
as as the brand that they need to
01:04:16
project.
01:04:17
>> Do you think any of them are more have a
01:04:20
stronger moral compass than others? cuz
01:04:22
I think Dario often gets the credit for
01:04:23
having more of a, you know, more of a
01:04:26
backbone and being more conscious of
01:04:28
implications.
01:04:31
>> He does get a lot of credit for that.
01:04:33
>> He's from Claude and Anthropic. For
01:04:34
anyone that doesn't know,
01:04:37
>> I don't think it truly matters that
01:04:41
question, the answer to that question,
01:04:43
because to me,
01:04:44
>> even if you were to swap all the CEOs
01:04:46
for someone that people would say is
01:04:49
better at running these companies, it
01:04:52
doesn't fix the problem that I identify
01:04:54
in the book, which is that there is a
01:04:56
system of power that has been
01:04:58
constructed where these companies and
01:05:00
the people running these companies get
01:05:02
to make decisions that affect billions
01:05:04
of people's lives. lives around the
01:05:05
world and those billions of people do
01:05:07
not get any say in how it goes.
01:05:10
>> Those people, they can go to the polls,
01:05:13
right? So, if the public are
01:05:14
sufficiently educated, they can go to
01:05:16
the polls and pick a leader that says
01:05:18
they're going to legislate or pass laws
01:05:21
or try and pass laws.
01:05:22
>> Yes.
01:05:23
But at the speed and pace at which these
01:05:26
companies operate and at the sheer scale
01:05:28
and size, they're able to also spend
01:05:31
extraordinary amounts of money, hundreds
01:05:33
of millions in this upcoming midterms to
01:05:35
try and kill every possible piece of
01:05:37
legislation that gets in their way and
01:05:39
craft legislation that would codify
01:05:40
their advantage.
01:05:42
And so to me,
01:05:45
I think sometimes as a society, we
01:05:47
obsess a little bit with
01:05:50
are these leaders good or bad people?
01:05:53
And to me the bigger question is is the
01:05:56
governance structure that we've created
01:05:59
a sound one or that allows broad
01:06:01
participation or an anti-democratic one
01:06:04
that has consolidated this
01:06:05
decision-making power in the hands of
01:06:06
the few because no person is perfect. It
01:06:09
does I don't I don't care who is on at
01:06:12
the top of these companies. they're not
01:06:14
going to have the ability to make
01:06:16
decisions on behalf of so many people
01:06:18
around the world who live and talk and
01:06:22
um and and have a culture and history
01:06:24
that are fundamentally different from
01:06:26
them without things going wrong.
01:06:29
And so that is why throughout history
01:06:31
we've moved from empires to democracy.
01:06:36
It's because empire as a structure is
01:06:39
inherently unound. it does not actually
01:06:42
maximize the chances of most people in
01:06:46
the world being able to live dignified
01:06:48
lives.
01:06:49
>> I'm going to try and take on their point
01:06:51
of view. So, this is me playing devil's
01:06:52
advocate. Okay. But Karen, if the US
01:06:58
don't continue to accelerate their
01:06:59
research with AI, at some point, China's
01:07:02
model is going to become so smart and
01:07:05
intelligent that we're basically going
01:07:07
to have to rent it off them and we're
01:07:08
going to be, you know, they'll get the
01:07:09
scientific discoveries. They'll discover
01:07:11
the new era of autonomous weapons and we
01:07:14
will be their backyard. And like
01:07:17
logically
01:07:19
that argument does appear to be pretty
01:07:21
true.
01:07:22
>> No, it's not.
01:07:23
>> If we scale up, if we just imagine any
01:07:25
rate of change with this intelligence,
01:07:26
at some point we're going to come to a
01:07:29
weapon that could theoretically disable
01:07:32
um all of the United States electricity,
01:07:34
their weapons systems. It would know
01:07:37
exactly how to disable the United States
01:07:39
from a cyber perspective because it
01:07:41
would be that smart. All you've got to
01:07:42
imagine is any rate of improvement of
01:07:44
any period any sort of long period of
01:07:46
time. So this is a theory that might be
01:07:50
true and if it's true
01:07:52
>> I mean yeah any theory might be true
01:07:55
>> but but if but but you know again going
01:07:57
to this point of like even if it's a
01:07:58
small percentage it's worth paying
01:07:59
attention to on the other side of the
01:08:00
foot. This is a theory that people talk
01:08:04
about. It could be the case that the
01:08:06
most intelligent civilization is going
01:08:09
to be the superior civilization.
01:08:12
Logically, that's a pretty sound thing
01:08:13
to say. No.
01:08:14
>> So, there's a lot of a lot of
01:08:17
fundamentals in this argument that would
01:08:19
need to be true in order for this to be
01:08:21
a viable argument. And let's knock them
01:08:23
down one by one. So the first one is
01:08:26
that
01:08:29
these systems are intelligent and that
01:08:31
just scaling them is going to bring us
01:08:32
more intelligence.
01:08:34
So far so true.
01:08:35
>> No, it's actually not because first of
01:08:39
all again we don't actually know if
01:08:42
these systems are like intelligence is
01:08:45
not it's not like the right analogy
01:08:46
almost. It's sort of like
01:08:50
it's like is a calculator a calculator
01:08:52
can do math problems faster than a
01:08:53
human. Does that make it intelligent?
01:08:56
>> It has a narrow intelligence because
01:08:57
they're solving a narrow problem which
01:08:58
is like 1 plus 1 equals 2. But
01:09:01
>> and these systems, they actually also
01:09:03
are quite narrowly intelligent in the
01:09:06
sense that even though these companies
01:09:07
say that they're everything machines
01:09:09
that can do anything for anyone, they
01:09:11
actually can only do some things for
01:09:12
some people. This is like the jagged
01:09:14
frontier of these AI models like some of
01:09:17
the capabilities are quite good, other
01:09:19
capabilities are not that good. You know
01:09:20
why that happens? is because the company
01:09:23
can only focus on advancing certain
01:09:24
types of capabilities. It can't
01:09:26
literally focus on advancing all types
01:09:28
of capabilities. They have to actually
01:09:30
set their mind to advancing a certain by
01:09:32
gathering the data that is needed for
01:09:33
that capability by taking uh you know
01:09:37
getting a bunch of human contractors to
01:09:39
annotate and train the model to do that
01:09:42
exact thing. And so
01:09:45
scaling these models is actually a
01:09:48
perpendicular question to are we
01:09:51
actually getting
01:09:53
more cyber capabilities specifically and
01:09:56
more military capabilities specifically.
01:09:58
>> I would argue that most of the most of
01:10:00
the top people in AI believe that the
01:10:02
intelligence is going to continue to
01:10:04
scale for some time. a lot of them do
01:10:06
like Jeffrey Hinton does.
01:10:07
>> And again, it's it's back to his
01:10:10
hypothesis about how human intelligence
01:10:12
works and what the appropriate model of
01:10:14
the brain is. His hypothesis throughout
01:10:17
his career has been the brain is a
01:10:19
statistical engine.
01:10:20
>> But that's his hypothesis and that is
01:10:22
not universally agreed upon especially
01:10:25
among people that are not in the AI
01:10:27
world. When you talk with
01:10:28
neuroscientists and psychologists,
01:10:29
people who actually study human
01:10:30
intelligence in the human brain, that is
01:10:32
where you start to get a lot of debate
01:10:35
and disagreement about this particular
01:10:36
view that Hinton has. And so this is
01:10:42
kind of like one of the one of the
01:10:44
things is like AI
01:10:46
is already being used in the military
01:10:48
and has been used in the military for a
01:10:50
long time. But ex specifically
01:10:54
accelerating large language models
01:10:57
isn't just the only path for getting
01:11:01
military cap. like the companies would
01:11:02
have to choose to specifically pick
01:11:05
military capabilities to accelerate not
01:11:08
just like general intell it's like you
01:11:10
know what I'm saying like they create
01:11:12
this myth that they are actually pushing
01:11:15
the frontier of all of the capabilities
01:11:17
of the model but that's not what's
01:11:18
actually happening internally and I have
01:11:20
I had hundreds of pages of documents on
01:11:22
like how they were specifically training
01:11:24
models they pick what capabilities they
01:11:27
want to advance and you know how they
01:11:28
pick them it's based on which industries
01:11:31
countries would be able to pay them the
01:11:32
most money for their services. So they
01:11:35
pick finance, law, medicine, healthcare,
01:11:40
commerce. It's not actually intelligent
01:11:43
like a like a a baby where you the the
01:11:47
more that you that the baby grows up,
01:11:48
they start having this like general
01:11:50
these general abilities.
01:11:52
>> I think I have jagged intelligence. I'll
01:11:54
be honest. I wasn't going to say it, but
01:11:57
I think I know a little I know a little
01:11:59
bit about uh No, I know a lot about a
01:12:01
little bit.
01:12:02
>> Yeah, but if but you also have the
01:12:04
capability to learn and acquire
01:12:05
knowledge by yourself. And you also have
01:12:06
the ability to choose what you're going
01:12:08
to learn and acquire by yourself.
01:12:10
>> It's not easy and it takes a lot more
01:12:11
time than these models. It seems less
01:12:13
compute, but
01:12:14
>> and you can learn how to drive in one
01:12:16
place and then immediately know how to
01:12:17
drive in another place. These models
01:12:19
cannot do that. Every time a
01:12:21
self-driving car is shifted to another
01:12:24
location, it has to completely retrain
01:12:26
on that location. It's like all the
01:12:28
self-driving cars. I mean, we're sitting
01:12:29
in Austin right now and there's all
01:12:30
these self-driving cars that are driving
01:12:32
through Austin.
01:12:34
But when one of them learns, they all
01:12:35
learn
01:12:36
>> which is which
01:12:37
>> well it's just because it's a it's an
01:12:40
operating system that is has an AI model
01:12:43
as part of it and you're training the AI
01:12:45
model and then you deploy that AI model
01:12:47
across all the self-driving
01:12:48
>> a big advantage because if one optimist
01:12:51
robot learns one thing in one factory
01:12:54
they all learn it and imagine that
01:12:56
imagine if humans if we all learned what
01:12:57
all the other humans learned that would
01:13:00
be that would give us such an
01:13:01
unbelievable competitive advantage. I
01:13:02
mean one of the ways we did that is
01:13:03
through communication.
01:13:04
>> They could not because they could be
01:13:05
learning the wrong thing which has also
01:13:06
happened again and again with these
01:13:08
technologies is that all of them then
01:13:10
learn the wrong thing and they all have
01:13:11
the same failure mode. I mean part of
01:13:13
the resilience of human society is that
01:13:15
we do have different expertises and we
01:13:17
also have different failure modes.
01:13:19
>> I think sometimes we hold AI models to a
01:13:21
higher standard than we hold humans to.
01:13:23
And in a weird because I I' I'd hear on
01:13:25
stage we're in we're in Austin at the
01:13:26
moment and I'd hear people go ah but you
01:13:29
know them AI models they hallucinate
01:13:30
sometimes. I'm like, "Have you met a
01:13:32
human?" Like, I I hallucinate all the
01:13:35
time. I can barely spell or do math.
01:13:39
>> So,
01:13:40
>> yes, but it's it's once again like using
01:13:42
this analogy that was specifically
01:13:43
picked in the early days of the field as
01:13:46
a way to market these technologies. like
01:13:48
we're repeatedly using the intelligence
01:13:50
analogy and relating these machines to
01:13:52
human intelligence as a a way to try and
01:13:56
gauge whether or not it is good or
01:13:59
worthy or capable in society. I think
01:14:01
the output is the thing that really m is
01:14:03
the most consequential which is like
01:14:04
okay it might have a different brain and
01:14:06
a different system but does it arrive at
01:14:07
the same capability like does it is it
01:14:10
able to do surgery on someone's brain is
01:14:12
it able to drive a car like my car
01:14:13
drives itself in in Los Angeles I don't
01:14:16
touch the steering wheel and I can drive
01:14:17
for many many hours and in here in
01:14:19
Austin I just saw the ones the other day
01:14:20
where they've removed the steering wheel
01:14:22
and the pedals the new cyber cabs so I
01:14:24
go it doesn't really matter if it's
01:14:25
using a different system if it's
01:14:26
navigating through the world as a car it
01:14:28
has a better safety record than human
01:14:30
beings
01:14:31
Um then as far as I'm concerned,
01:14:34
intelligence or not, it's like
01:14:36
>> yes, you know,
01:14:36
>> but that was not the original argument
01:14:38
that you made, which was like these
01:14:40
systems are just generally going to
01:14:41
become more intelligent across different
01:14:43
things based on the prediction. This is
01:14:46
a prediction that you're making, right?
01:14:47
Like that and this is a prediction that
01:14:49
all the AI um
01:14:50
>> Ilia's making, Dario's making, Elon's
01:14:52
making, Zuckerberg's making, man's
01:14:54
making, Dennis is making.
01:14:56
>> And do you know what the common feature
01:14:57
of all of them is? They profit
01:14:59
enormously off of this myth.
01:15:01
>> Elon has recently spearheaded the
01:15:04
construction of Colossus, a massive
01:15:05
supercomputer in Memphis housing a
01:15:07
100,000 GPU specifically to scale up
01:15:10
their API models faster than their
01:15:12
competitors. It appears that they've all
01:15:14
converged around this idea that you can
01:15:16
brute force your way to greater, more
01:15:18
generalized intelligence. They've
01:15:20
converged around the idea that you can
01:15:22
brute force your way into models that
01:15:24
they can sell to people for automating
01:15:27
certain tasks that are that are
01:15:29
financially lucrative.
01:15:30
>> And I heard Elon say that if you're a
01:15:32
surgeon, there's just no point. He was
01:15:33
like, don't train to be a surgeon. He
01:15:35
says in a couple of years time, Optimus
01:15:37
and AI generally are going to be better
01:15:39
than any surgeon that's ever lived.
01:15:40
>> Yeah. You know,
01:15:41
>> do you think these things are true?
01:15:42
Well, you know, I I'm pretty sure it was
01:15:44
Hinton that famously slash infamously
01:15:46
said there would be no need for
01:15:48
radiologists anymore.
01:15:50
>> There would be no need for radiologists
01:15:51
anymore in he set a deadline that we've
01:15:54
already passed. I don't remember how
01:15:56
many years.
01:15:58
Radiology is doing great as a
01:16:00
profession.
01:16:00
>> Do you think it will be in 5 years?
01:16:02
>> Okay. So, this this once again goes back
01:16:05
to this question of like why do we build
01:16:06
technology and why should we
01:16:08
specifically be building AI? Okay. And
01:16:11
for me like the whole project of
01:16:14
technology development advancement is
01:16:15
not to advance technology for
01:16:17
technologies sake.
01:16:18
>> It's to help people.
01:16:21
And there have been lots of research
01:16:23
that has shown that actually the best
01:16:26
outcomes for people in a healthcare
01:16:28
setting is for the radiologist to have
01:16:31
the AI model in their hands
01:16:36
and for the for the human expert to use
01:16:40
the AI model as a tool as an input into
01:16:43
their judgment. And it is that
01:16:45
combination that leads to the most
01:16:48
accurate and early diagnoses of certain
01:16:51
types of cancer that then help improve
01:16:53
the prognosis of the patient.
01:16:55
>> Do you believe that in the coming years
01:16:58
all the cars pretty much all the cars on
01:16:59
the road will be driving themselves?
01:17:00
>> No.
01:17:01
>> You don't you don't think so?
01:17:02
>> Mm-m.
01:17:02
>> How come?
01:17:03
>> Because of the way the technology works.
01:17:06
>> Because because these are statistical I
01:17:09
mean currently the way that AI models
01:17:11
are primarily developed. They're
01:17:13
statistical engines. You have what's
01:17:15
called a neural network, which is a
01:17:17
piece of software that has a bunch of
01:17:20
densely connected nodes and
01:17:22
>> like parameters. Is this what they call
01:17:24
parameters?
01:17:24
>> Yeah, pretty much. And you're just
01:17:26
pumping a bunch of data into it and then
01:17:29
it's analyzing the data and creating
01:17:31
this all of these finding all these
01:17:33
correlations in the data, finding all
01:17:34
these patterns and then it's through
01:17:36
those patterns that the machine is then
01:17:39
able to act autonomously, right? And so
01:17:42
the way that they're training a
01:17:43
self-driving car is they're they're
01:17:45
recording all this footage and then they
01:17:48
have tens of thousands or hundreds of
01:17:49
thousands of human contractors that draw
01:17:53
literally around every single vehicle in
01:17:57
the footage, every single pedestrian,
01:18:00
every single traffic light, every single
01:18:02
lane marking and label it exactly as
01:18:04
such. So that then it's fed into an AI
01:18:07
model that can identify all of these
01:18:10
different components and then it's
01:18:11
connected to another piece of software
01:18:14
that is not AI that's saying okay if you
01:18:17
if the AI model recognizes the
01:18:19
pedestrian we do not run over the
01:18:21
pedestrian.
01:18:23
If the AI model recognizes a red traffic
01:18:26
light we stop. And so the like the thing
01:18:30
about statistical engines is that it's
01:18:32
based on probabilities. It's not based
01:18:34
on deterministic logic.
01:18:37
So
01:18:39
systems make errors all the time and
01:18:41
it's impossible. It is technically
01:18:44
impossible to get them to stop making
01:18:47
errors.
01:18:48
>> Humans make errors way more than
01:18:50
>> systems in this case. Like the safety
01:18:53
record is like isn't it like 10 times
01:18:54
more safe to be driven in a Tesla with
01:18:57
autonomous driving than it is to for a
01:18:59
human to drive?
01:18:59
>> It depends on the place. It depends on
01:19:02
whether the Tesla was trained to
01:19:03
specifically navigate the place that
01:19:05
you're driving.
01:19:05
>> Get drunk
01:19:06
>> because if it's in Mumbai,
01:19:09
>> in some place in Vietnam, no, it would
01:19:12
not be safer. I WOULD MUCH RATHER be
01:19:15
driven
01:19:16
>> by someone that has been driving in that
01:19:18
place their whole life. I'm I'm not
01:19:20
arguing against like the fact that in
01:19:22
certain places where the car has been
01:19:24
explicitly trained to drive in this
01:19:26
place that it has a better safety record
01:19:29
than the humans that are driving in that
01:19:30
place. But you specifically asked if I
01:19:33
think that all of the
01:19:34
>> most cars
01:19:35
>> most cars in the world in the US
01:19:38
>> in the United States cuz we're here.
01:19:40
>> I don't actually think that it's like
01:19:41
imminently on the horizon
01:19:43
>> 10 years.
01:19:44
>> No, I don't think so.
01:19:45
>> I sat with Dra from Uber and he's pretty
01:19:47
convinced that his 9 million couriers
01:19:48
will be replaced by autonomous vehicles.
01:19:51
>> I mean, how long have has self-driving
01:19:53
cars been
01:19:55
invested in thus far? It's been more
01:19:57
than 10 years. And what percentage of
01:19:59
cars right now are autonomous
01:20:03
>> on the US roads? I mean, so part of it
01:20:05
is it's actually not a technical
01:20:07
problem, right? Like part of it is also
01:20:09
social problem like do people even trust
01:20:11
getting into these vehicles? Part of it
01:20:13
is also a legal problem which is if the
01:20:16
car the self-driving car kills someone,
01:20:19
which it has happened.
01:20:20
>> Yeah, it has happened.
01:20:22
>> Who is responsible? So, in the case in
01:20:24
LA, it was both Tesla and the driver
01:20:26
because the driver dropped their phone,
01:20:29
they looked down, and this was a couple
01:20:30
of years ago, I believe. Um, and they
01:20:32
went to grab their phone and they hit
01:20:34
someone, and so it went to court, and
01:20:36
they were held both responsible, both
01:20:38
the driver and Tesla. Um, in terms of
01:20:42
Tesla,
01:20:44
pretty much everyone that gets the car,
01:20:46
it comes with autonomy now for pretty
01:20:47
much most people, I believe.
01:20:49
>> Partial autonomy. Yeah, it's called full
01:20:50
self-driving at the moment where it's
01:20:51
like
01:20:52
>> I mean, yes, it is called full
01:20:53
self-driving.
01:20:54
>> Full self-driving supervised where you
01:20:56
kind of have to be looking in the d. You
01:20:57
have to be looking in the right
01:20:58
direction, but
01:20:59
>> Yeah. So, it's partial autonomy.
01:21:01
>> And here in Austin, it's full autonomy
01:21:04
cuz there's no steering wheel.
01:21:05
>> Yeah.
01:21:06
>> On the new car. Um, so you can't drive
01:21:07
it anyway. But it is, you know, the
01:21:09
Model Y is the undisputed highest
01:21:11
selling car, bestselling car in the
01:21:13
world across all brands. Well, I guess
01:21:16
my point here is like these predictions
01:21:19
where they say AI is going to completely
01:21:22
change transportation and driving. It's
01:21:24
going to completely change lawyers
01:21:25
aren't going to have jobs. Accountants
01:21:26
aren't going to have jobs. Um, do you
01:21:29
believe that they are true? Do you
01:21:30
believe that there's going to be mass
01:21:31
job displacement?
01:21:33
>> Okay, so I do think that there is going
01:21:35
to be huge impacts on employment and we
01:21:37
already seeing those impacts.
01:21:39
It is not simply because the AI models
01:21:42
are just automating those jobs away. It
01:21:44
is specifically
01:21:47
because the models are improving in
01:21:49
certain capabilities based on what the
01:21:51
companies that are developing them
01:21:52
choose to improve them on. And
01:21:56
executives at other companies are then
01:21:58
deciding to fire or lay off their
01:22:00
workers because they think that AI can
01:22:04
replace the worker irrespective of
01:22:06
whether that might be true. And there,
01:22:08
you know, there have been cases of like
01:22:09
the CLA CEO who laid off a bunch of
01:22:11
people thinking that he would replace
01:22:12
everyone with AI and then it didn't
01:22:13
actually work and he had to ask some
01:22:15
people to come back.
01:22:16
>> I actually DM'd him about this. If
01:22:18
you're hearing this, this is because
01:22:19
I've DM'd Sebastian and he's fine with
01:22:21
me sharing this.
01:22:22
>> He said, because I've heard his name
01:22:24
mentioned a lot and so when I when we
01:22:25
talked about AI in the past and people
01:22:27
mention Sebastian and Cler as the
01:22:29
example, I wanted to clarify with him
01:22:31
what the truth was.
01:22:32
>> He said, "It's great to hear from you.
01:22:33
Um, I think sometimes people struggle
01:22:35
with two things can be true at the same
01:22:37
time. I think it might be time to come
01:22:39
back on your podcast.
01:22:41
To your point, this is the media
01:22:42
misinterpreting my tweet. We are
01:22:44
doubling down on AI more than ever. Cler
01:22:46
is shrinking with almost 100 employees
01:22:48
per month due to AI. We used to be 7,400
01:22:52
at the peak. A year ago, 5,500. Now
01:22:56
we're 3,300.
01:22:58
And by the end of summer, so this was
01:23:00
last year, will be 3,000 people. AI
01:23:04
handles 70% of our customer service
01:23:06
conversations at this moment. This is
01:23:08
because we have realized that with AI,
01:23:10
the production cost of software comes
01:23:12
down to almost zero. Just like
01:23:13
manufacturing used to be all handcrafted
01:23:15
and then the machines came. Code used to
01:23:17
be all handcrafted up until a few years
01:23:19
ago. And now it is machine produced. And
01:23:23
ultimately we pay people more than ever
01:23:26
for the unique handcrafted man-made
01:23:29
stuff. China is a bank. People will want
01:23:31
to connect to humans not only machines.
01:23:33
They want us to be personable,
01:23:35
relatable, even flawed. So we need to
01:23:38
make sure while we are automating
01:23:40
replacing with AI in parallel, we make
01:23:42
sure we offer a super available human
01:23:46
experience. I'm really glad you read
01:23:48
this because I think it touches on some
01:23:50
really important nuances to
01:23:54
the AI. Yeah. Like the impact that AI is
01:23:56
going to have on employment. So I think
01:23:58
the there's often these binary
01:24:00
narratives. It's like AI is going to
01:24:02
come for every job.
01:24:04
>> Mhm.
01:24:04
>> Or people say AI is not actually working
01:24:07
and it's not actually coming for jobs.
01:24:09
And like the reality is it's coming for
01:24:11
jobs. There are definitely jobs that are
01:24:14
being automated away because of the
01:24:17
capabilities of their models. And
01:24:18
there's also jobs that are being lost
01:24:19
because executives are deciding to lay
01:24:21
off the workers even if the models don't
01:24:23
match the capabilities because it's good
01:24:25
enough. Like they would rather have the
01:24:26
good enough model for way cheaper
01:24:28
>> or they made a mistake with hiring. They
01:24:30
blowed their team and it's a great
01:24:31
convenient thing to say.
01:24:32
>> Exactly. Like there's there's there's
01:24:34
many reason but like clearly we're
01:24:35
already seeing impacts on the job
01:24:37
market. Like the um US jobs report that
01:24:40
came out earlier this year showed that
01:24:43
there has been a decline in hiring is a
01:24:47
slowdown in hiring across especially
01:24:49
white collar professional industries.
01:24:53
And you saw Anthropic's report the new
01:24:54
this week. The TLDDR is it matches kind
01:24:56
of what you were saying where they
01:24:57
Anthropic looked at exactly how people
01:25:00
were using their models and they looked
01:25:02
at like what people are saying.
01:25:04
>> Yeah.
01:25:04
>> And they said that there's been a 40%
01:25:06
reduction in entry- level jobs in
01:25:08
particular and then they made this graph
01:25:09
which has gone viral over the internet.
01:25:11
The red shows where we are now in terms
01:25:12
of capability and based on how people
01:25:15
are currently using the models they
01:25:17
prediction
01:25:17
>> extrapolated out that the blue part will
01:25:19
be the disrupted parts. This is the
01:25:21
things that they say AI can do right
01:25:23
now, but people don't realize it yet.
01:25:25
So, if you look at it, it's like it's
01:25:27
kind of all the stuff you would expect.
01:25:28
>> Yeah.
01:25:29
>> It's the physical real world human stuff
01:25:31
>> which robots maybe can do someday like
01:25:33
construction or agriculture that are
01:25:35
untouched, but like office and admin, um
01:25:38
like saying finance stuff, math,
01:25:40
>> and notice that these are all the things
01:25:42
that I just named that they purposely
01:25:44
>> finance, math, law,
01:25:46
>> media and arts. That's me cooked.
01:25:48
>> Yeah.
01:25:50
office and admin. I mean they do focus a
01:25:52
lot on like assistant type and
01:25:55
managerial work.
01:25:56
>> So but but the the other thing that the
01:25:59
CLO CEO said was
01:26:02
but people also want human experiences.
01:26:05
So it's not actually just about the
01:26:07
capabilities of the models. It's also
01:26:08
about what people want like some things
01:26:11
they would turn to AI for and some
01:26:14
things they wouldn't irrespective of
01:26:16
whether or not AI is capable of doing it
01:26:19
but because of a preference that they
01:26:22
want humanto human interaction
01:26:24
>> and so what we're seeing right now is
01:26:29
yeah the the thing that happens with
01:26:30
every wave of automation which is that
01:26:33
there is a bunch of entry-level work
01:26:34
that gets automated away and there There
01:26:38
are also new jobs created, but the jobs
01:26:40
that are created are one in one of two
01:26:42
categories. There are people that get
01:26:45
even higher skilled jobs and what he was
01:26:47
saying like we pay people more for like
01:26:49
the handcrafted code now
01:26:52
>> and there's also the people who get way
01:26:54
worse jobs and so there was this amazing
01:26:57
article in New York magazine that was
01:26:59
talking about how a lot of people are
01:27:02
getting laid off and then they end up
01:27:05
working in data annotation which is the
01:27:09
labor that I've been referring to
01:27:10
throughout this conversation that
01:27:11
companies need in order to teach their
01:27:14
models the next thing that the companies
01:27:16
are trying to automate. And so like a
01:27:18
marketer gets laid off and then they go
01:27:21
and work for a data annotation firm to
01:27:24
train the models on the very job that
01:27:27
they were just laid off in which will
01:27:29
then perpetuate
01:27:31
more layoffs if that model then develops
01:27:33
that skill. And the article was talking
01:27:37
about how this has become a huge
01:27:43
catchall for a lot of people that are
01:27:44
struggling with finding job
01:27:46
opportunities right now, including like
01:27:48
awardwinning directors in Hollywood that
01:27:51
are actually secretly doing this data
01:27:52
annotation work to put food on the
01:27:54
table. And so when they talk about
01:27:58
there's going to be mass unemployment
01:28:01
and then there's going to be some new
01:28:02
jobs created that we can't even imagine,
01:28:04
I think a lot of these narratives rarely
01:28:06
talk about like first of all, why are
01:28:09
some jobs going away? It's not just
01:28:10
because of the model capabilities, it's
01:28:11
also because of executive choices and
01:28:13
because of the rhetoric that they use if
01:28:15
they want to just downsize. Um, but the
01:28:18
other thing that is rarely talked about
01:28:20
is the jobs, a lot of the jobs that are
01:28:22
created are way worse than the jobs that
01:28:26
were there
01:28:27
>> and it breaks the career ladder. So,
01:28:29
it's the entry level and the mid tier
01:28:31
jobs that get gouged out. It's higher
01:28:34
order jobs and then way more lower order
01:28:37
jobs that get created. And so, how do
01:28:42
people continue to progress in their
01:28:44
careers? There's no more rungs on the
01:28:45
ladder.
01:28:46
>> I actually don't know the answer to this
01:28:47
question. And I've been furiously trying
01:28:48
to find a good answer to this question
01:28:50
because I can, you know, everything is
01:28:52
theory. And for my audience, I would say
01:28:55
most of my audience don't run
01:28:56
businesses. A lot of them do, a lot of
01:28:58
them aspire to, but they don't run
01:28:59
businesses. So, they're kind of, they're
01:29:00
also in the land of theory. They're
01:29:02
hearing lots of different things. Jack
01:29:03
Dorsey does his tweet saying he's
01:29:04
halfing his headcount because of AI.
01:29:06
They don't know what's true. They don't
01:29:07
know the sort of internal economics at
01:29:09
Jack's company and did he bloat the
01:29:11
company during the pandemic and he's
01:29:12
just using this as an excuse to make
01:29:13
this share price spike seven points
01:29:15
because his investors now think they're
01:29:16
an AI company or whatever.
01:29:18
>> Mh.
01:29:18
>> It's hard to pass through. So eventually
01:29:20
I go, okay, what am I doing?
01:29:22
>> I have hundred hundreds of team members,
01:29:24
probably 70 companies I invest in, maybe
01:29:26
five or six that I'm like the lead
01:29:27
shareholder in. What am I actually doing
01:29:29
on a day-to-day basis right now? I am
01:29:31
I'm also I also consider myself to be
01:29:32
head of recruitment
01:29:34
>> but in the last month in particular I
01:29:36
have met extremely capable candidates in
01:29:38
terms of cultural alignment hard work
01:29:40
those kinds of things but I've had to
01:29:42
take a great deal of pause because when
01:29:44
I run the experiment of can I get an AI
01:29:46
agent to do that exact same thing the
01:29:48
answer is increasingly yes
01:29:50
>> especially in a world of open clause
01:29:53
>> and so what I'm curious like
01:29:56
>> now you confront this decision where
01:29:58
you're seeing in this short-term period
01:30:01
you could just choose the AI agent
01:30:05
and in the long-term period
01:30:07
there is no career ladder. So, so who
01:30:11
are you promoting into these senior
01:30:13
roles? Like what how do you resolve it
01:30:15
for your own company?
01:30:16
>> Yeah, it's a good question. So, there's
01:30:17
kind of two ways I'm thinking about it.
01:30:18
I think really deep expertise is very
01:30:21
very valuable because if you're now the
01:30:22
orchestrator of potentially AI agents,
01:30:25
it's really about um having a deep
01:30:27
understanding of the right question to
01:30:28
ask and and that's someone who has deep
01:30:30
expertise on something. So I need my CFO
01:30:33
>> because if she's going to be
01:30:34
orchestrating our team of agents that
01:30:36
might be doing financial analysis or
01:30:37
whatever else, she needs to understand
01:30:40
what to tell them to do in our company.
01:30:43
>> Mhm.
01:30:43
>> And in turn financial analysts can't do
01:30:45
that. They need this the 50 odd years of
01:30:47
experience that you know CLA has. On the
01:30:50
other end, I need Cass. Cass is 25. Cass
01:30:53
knows everything about AI agents. He's a
01:30:56
young Japanese kid who's highly highly
01:30:58
curious. You know, on the weekend, he's
01:30:59
building AI agents to solve problems in
01:31:01
my life. I need those two kinds of
01:31:03
thinking, which is highly proficient
01:31:06
agent maxing young kids or they don't
01:31:07
necessarily need to be young, but like
01:31:09
really lean in high curiosity. That's
01:31:11
creating a force multiplier in my
01:31:12
business. And then I need deep
01:31:13
expertise. Now the everything else
01:31:16
outside of there is another one I've
01:31:18
thought of another group is like people
01:31:19
with extremely great IRL people skills
01:31:23
>> because we do meet people in real life.
01:31:26
We greet you when you arrive here. We
01:31:27
greet we when we go for lunch with big
01:31:29
clients that we have whether it's Apple
01:31:31
or LinkedIn or whoever it might be. We,
01:31:32
you know, we need to smoosh.
01:31:34
>> Mhm.
01:31:35
>> And we have teams who, you know, are in
01:31:37
person in the office. So, we we do a lot
01:31:39
of stuff IRL and increasingly we're
01:31:41
building communities even for this show.
01:31:42
We're doing community events all around
01:31:43
the world. So, we need people that are
01:31:44
good at that as well. IRL, bringing
01:31:47
people together in real life and
01:31:48
organizing stuff. Those are the three
01:31:49
groups of people that I'm like, you
01:31:51
know, irreplaceable right now. And if
01:31:54
you were to to all of the all the roles
01:31:58
that could be done by AI agents, if we
01:32:00
were to replace them with AI agents, do
01:32:01
you think you would still have these
01:32:02
three roles pools of people to hire and
01:32:06
promote into the three critical things
01:32:08
that you need in the long term?
01:32:10
>> If things carry on at the the current
01:32:12
rate of trajectory,
01:32:14
>> yeah,
01:32:14
>> one could assert that even those roles
01:32:16
would experience pressure. If you just
01:32:18
imagine like people think of things
01:32:20
either statically or linearly or
01:32:21
exponentially. Yeah,
01:32:22
>> you imagine an exponential rate of
01:32:24
improvement, which is kind of what I've
01:32:25
seen. Even like a 10% compounding rate
01:32:26
of improvement at some point,
01:32:32
>> at some point, at some point, I think
01:32:34
what remains is actually the IRL
01:32:37
irreplaceably human stuff, human to
01:32:39
human, our Maslovian needs of being in
01:32:41
person like we are now aren't going to
01:32:43
change. We need connection. Humans get
01:32:44
very sick when they don't have other
01:32:46
human beings in their life and strong,
01:32:48
deep relationships. 100% agree. So that
01:32:51
stuff is going to matter a whole lot. I
01:32:53
have this contrarian weird take that
01:32:54
actually maybe this is the first
01:32:56
technology that's going to deliver on
01:32:57
the promise of making us human and
01:32:59
connected because we're going to be
01:33:00
rendered useless of everything else
01:33:01
other than what humans are good at. Cuz
01:33:04
all the other technology said, "Oh,
01:33:05
we're going to make you more connected,
01:33:06
connecting the world." And they
01:33:08
disconnected the world and isolated the
01:33:09
world. But maybe this is the one. It's
01:33:10
so intelligent now that it doesn't need
01:33:12
us to [ __ ] around in spreadsheets
01:33:13
anymore.
01:33:13
>> Do you see
01:33:16
that actually happening in real time
01:33:18
right now that it's making us more
01:33:20
able to be in person, connected with one
01:33:23
another, having deeper social community
01:33:26
engagements.
01:33:28
>> Yes.
01:33:29
>> Yes.
01:33:29
>> And I'll give you some data points.
01:33:31
>> Okay.
01:33:31
>> Data point number one, the Financial
01:33:33
Times released a report on social media
01:33:35
usage. And what they saw is 2022 was the
01:33:39
peak and it's plateaued ever since. The
01:33:40
generation that's plateaued the fastest
01:33:42
and heading down is the younger
01:33:44
generations. The boomers are still off
01:33:46
to the races, right? So on Facebook and
01:33:48
stuff. And then you look at the way Gen
01:33:50
Alfa are using social media. They're not
01:33:52
posting as much. They call it uh posting
01:33:54
zero. They're scrolling sometimes, but
01:33:56
they're in dark social environments like
01:33:57
WhatsApp and Snapchat and iMessage.
01:33:59
They're not like performing to the
01:34:00
world. They also value IRL experiences
01:34:02
much more than any other generation.
01:34:04
They're like not getting smashed. We're
01:34:05
seeing every brand has a run club.
01:34:08
um I mean runs exploding around the
01:34:10
world and we're seeing this real sort of
01:34:13
sort of almost like innate realization
01:34:16
that like technology let us down at some
01:34:18
fundamental level like dating apps let
01:34:20
us down social networking kind of has
01:34:22
let us down and we're seeing I think
01:34:24
maybe a bifocation of society where a
01:34:26
lot of people are going [ __ ] this like I
01:34:27
want to go back to what it is to be a
01:34:29
human
01:34:29
>> and I I would imagine that in such a
01:34:31
world where intelligence is so
01:34:33
sophisticated that we no longer needed
01:34:34
to sit at laptops and like I think
01:34:37
screen time is going to continue to
01:34:38
fall. I think you go into an office,
01:34:39
you're not going to see people sat at
01:34:40
laptops. You're gonna see something
01:34:41
completely different. And I think maybe,
01:34:45
you know, and then we talk about robots
01:34:47
and Optimus robots. Elon says there'll
01:34:48
be 10 billion Optimus robots. Elon has
01:34:51
been wrong with timing before. He's
01:34:54
almost never been wrong on the big
01:34:56
things completely. He's just his timing
01:34:59
is got a bad track record. Um, so I
01:35:02
think he's he's probably right. You
01:35:03
know, I think I've I've got some people
01:35:05
on the way from Boston Dynamics and
01:35:06
these other big companies like Scale AI,
01:35:08
and they're actually bringing the robots
01:35:09
here to show it, like folding laundry,
01:35:11
doing the dishes. I'm not saying that's
01:35:12
what I would want in my home, but I
01:35:13
think factory work is going to
01:35:15
completely change. I think a lot of
01:35:16
manual labor is going to completely
01:35:17
change, and I think we're going to be
01:35:18
forced to do what only we can do. Um,
01:35:22
Sebastian, who's the CEO of Cler, has
01:35:24
actually just called me.
01:35:29
>> Hello, Sebastian. You're right.
01:35:30
>> Hey, how are you?
01:35:31
>> I'm good. How are you?
01:35:33
It's been a while.
01:35:34
>> It has been a while since you're on the
01:35:36
show. I was just saying we do need to
01:35:37
get you back on.
01:35:38
>> I I just I just had a couple of simple
01:35:40
questions cuz you know I do a lot of
01:35:41
interviews and um Clan has always
01:35:43
mentioned because I think the media has
01:35:45
said that you like double down on AI
01:35:46
then you reversed because it didn't work
01:35:48
out. So I know I spoke to you a while
01:35:50
ago and we exchanged a couple of DMs
01:35:51
about it but that was more than a it was
01:35:53
almost a year ago now.
01:35:54
>> So I just wanted to get an update on
01:35:56
Cler's business AI agents and all of
01:35:58
that if possible. First and foremost, we
01:36:00
were early on uh released um AI uh to
01:36:04
support our customer service which had
01:36:06
that uh initial uh benefit of uh more
01:36:10
calls being dealt with by AI which
01:36:12
customers liked because those calls or
01:36:13
chat messages were much much faster and
01:36:16
more qualitative. Then since then that
01:36:18
has actually expanded slightly. Um what
01:36:21
we did however try to communicate as
01:36:23
well is that we believed in a world of
01:36:25
where AI is cheap and available the
01:36:28
value of human interaction will be
01:36:31
regarded as higher. So the future of
01:36:33
customer service VIP is a human um we
01:36:37
have then hence doubled down on
01:36:38
providing more of that but at the same
01:36:41
time the efficiency gains within the
01:36:42
company has continued. I mean we used to
01:36:45
be about 6,000 people and and now we are
01:36:49
less than 3,000 which is 2 3 years since
01:36:52
we stopped recruiting and at same point
01:36:54
in time our revenue has doubled right so
01:36:57
you can clearly see that AI has allowed
01:36:59
us to be do more with less people but we
01:37:02
have avoided layoffs and instead relied
01:37:05
on natural attrition when people kind of
01:37:08
move on to other jobs. I mean from my
01:37:11
perspective we will continue to be very
01:37:14
you know not really recruit much. I mean
01:37:16
we recruit a little bit here and there
01:37:17
but we expect that kind of natural
01:37:19
attrition of 10 15% per year to continue
01:37:23
and to become fewer. I think the big
01:37:26
breakthrough was really in November
01:37:27
December last year where even the kind
01:37:30
of more most skeptical
01:37:33
uh engineers who were like very
01:37:35
well-renowned and and appreciated like
01:37:37
the founder of Linux and stuff like that
01:37:39
basically said that coding has now been
01:37:42
resolved and hence is not you know uh
01:37:45
you don't need to code anymore and that
01:37:46
was kind of a common sentiment. So I
01:37:48
think in in coding that's definitely an
01:37:51
engineering work that has been a
01:37:53
tremendous shift in the last six months.
01:37:55
>> What do all these people go do
01:37:57
Sebastian?
01:37:58
>> I am optimistic. I mean I think
01:38:00
obviously people will have a lot of
01:38:02
opinions about this topic but I still
01:38:05
believe that we are going to move
01:38:07
towards a richer society. Now in the
01:38:09
short term there could be more worry
01:38:12
about what happens if people don't get a
01:38:14
job and and so forth. But I think in the
01:38:16
longer term, I I am optimistic what it
01:38:19
means for society and humanity.
01:38:21
>> Thank you so much, Seb. I'll chat to you
01:38:23
soon. Thank you for taking the time. I
01:38:24
appreciate you, mate. Thanks.
01:38:25
>> All right. All right. Byebye. Byebye.
01:38:28
>> You know the little traditional SIM card
01:38:30
that goes inside of our phones. They
01:38:32
haven't changed at all since they were
01:38:34
invented in the '90s. You have this
01:38:36
physical piece of plastic that means
01:38:37
you're locked into one carrier, one
01:38:39
network, and the second you cross a
01:38:42
border, that carrier can start charging
01:38:43
you whatever they want. But there are
01:38:45
alternatives and today's sponsor SY is
01:38:48
one of them. It's an eSIM app that gives
01:38:50
you a safe and secure data connection in
01:38:53
over 200 destinations. All of their
01:38:55
ESIMs have built-in cyber security which
01:38:57
is great if you're traveling for work
01:38:58
and looking at confidential material.
01:39:00
I've been using SY whenever I travel
01:39:02
because the connection is always
01:39:03
reliable and it saves me a ton of
01:39:05
roaming fees. It also means I don't have
01:39:07
to deal with all of the faf that
01:39:09
surrounds sorting out a SIM everywhere I
01:39:11
go. If you want to give it a try,
01:39:12
download the sale app from the app store
01:39:14
now and scan the QR code on screen. And
01:39:16
if you want 15% off your first purchase,
01:39:19
use my code D O A when you get to check
01:39:23
out. That's D O A for 15% off. Keep that
01:39:28
to yourself. This is something that I've
01:39:30
made for you. I've realized that the Dio
01:39:32
audience are strivals
01:39:36
that we want to accomplish. And one of
01:39:38
the things I've learned is that when you
01:39:40
aim at the big big big goal, it can feel
01:39:43
incredibly psychologically uncomfortable
01:39:46
because it's kind of like being stood at
01:39:47
the foot of Mount Everest and looking
01:39:49
upwards. The way to accomplish your
01:39:51
goals is by breaking them down into tiny
01:39:54
small steps. And we call this in our
01:39:56
team the 1%. And actually this
01:39:57
philosophy is highly responsible for
01:40:00
much of our success here. So, what we've
01:40:02
done so that you at home can accomplish
01:40:04
any big goal that you have is we've made
01:40:06
these 1% diaries and we released these
01:40:09
last year and they all sold out. So, I
01:40:12
asked my team over and over again to
01:40:13
bring the diaries back, but also to
01:40:14
introduce some new colors and to make
01:40:16
some minor tweaks to the diary. So, now
01:40:18
we have a better range for you. So, if
01:40:22
you have a big goal in mind and you need
01:40:24
a framework and a process and some
01:40:26
motivation, then I highly recommend you
01:40:28
get one of these diaries before they all
01:40:30
sell out once again. And you can get
01:40:32
yours at the diary.com.
01:40:34
And if you want the link, the link is in
01:40:36
the description below.
01:40:38
>> Any thoughts? Well, I actually had
01:40:40
thoughts on something that you said
01:40:42
before he called,
01:40:44
>> which is you were saying that the
01:40:46
Jenzers like there's this trend that
01:40:48
they're actually disconnecting from
01:40:49
technology. So, they're becoming more in
01:40:51
person. And then there's this other
01:40:53
class of workers that are actually
01:40:54
leaning into the technology, but then
01:40:56
becoming more human because they're
01:40:57
leaning into the technology
01:41:00
>> because they're realizing that they
01:41:01
should actually just be spending more
01:41:03
time doing inerson interactions rather
01:41:06
than staring at a spreadsheet. And so
01:41:08
they're no longer doing the typing,
01:41:09
whatever. I really want to go back to
01:41:10
this New York Magazine piece that just
01:41:12
came out
01:41:13
>> because what you're describing is true
01:41:16
for a very specific category of people,
01:41:18
which is often like the business owners
01:41:20
and leadership within companies that
01:41:22
actually can make these decisions on how
01:41:25
they spend their time and what they
01:41:27
ultimately do with their time. But what
01:41:30
the piece talks about is the working
01:41:34
class like people like people who are
01:41:36
not business owners that are then having
01:41:39
to experience being laid off and then
01:41:43
working for the data annotation industry
01:41:46
which is now one of the top jobs on
01:41:48
LinkedIn by the way. Um the yeah so
01:41:51
LinkedIn had a report that showed the
01:41:53
top 10 jobs with the highest growth in
01:41:56
the last year and data annotation is on
01:41:59
that list.
01:42:00
>> And for anyone that doesn't know what
01:42:01
data annotation is.
01:42:02
>> Yeah. So data annotation is the process
01:42:05
of teaching these chat bots or or any AI
01:42:09
system to do what they ultimately are
01:42:12
able to do. So the fact that chat GBT
01:42:14
can chat is because there were tens of
01:42:16
thousands or hundreds of thousands of
01:42:17
people that were literally typing into a
01:42:20
large language model and showing it.
01:42:23
This is how you're supposed to then
01:42:24
respond when a user types in a prompt
01:42:27
like this. Before they did that work,
01:42:31
chatgbt didn't exist. Like it just it
01:42:34
would just you would prompt the model
01:42:35
and the model would generate some text
01:42:37
that was not in dialogue with the
01:42:39
person. It would kind of generate
01:42:40
something that was adjacently related.
01:42:42
Is this what they call reinforcement
01:42:43
learning where you kind of you give it
01:42:44
like a
01:42:45
>> it's a part of the process of
01:42:46
reinforcement learning. So you do data
01:42:48
annotation which is literally um showing
01:42:51
lots of different
01:42:53
um you know examples of things that you
01:42:55
want the model to know and then
01:42:57
reinforcement learning is getting the
01:42:58
model to then train on those examples
01:43:00
iteratively in a way that then
01:43:02
>> gives the model some of those
01:43:04
capabilities. And what the New York
01:43:07
Magazine piece highlighted is many many
01:43:10
of the people that are getting laid off
01:43:12
now or or or are struggling to find
01:43:14
work. And these are highly educated
01:43:16
people. They're college graduates, PhD
01:43:19
graduates, law degree graduates,
01:43:21
doctors, um and again like award-winning
01:43:24
directors that are that are then
01:43:27
struggling to find employment in the
01:43:29
economy because the economy has been
01:43:31
very much restructured by AI. they are
01:43:33
then finding themselves being serving
01:43:36
this industry and the industry is
01:43:39
designed in a way that is extremely
01:43:41
inhumane because what the companies the
01:43:45
companies that use these data annotation
01:43:47
services like there's these third party
01:43:48
providers that are data annotation firms
01:43:52
an open AI a gro um a Google they will
01:43:55
hire these firms to then find the
01:43:58
workers to perform the data annotation
01:44:00
tasks that they need for these These
01:44:02
firms, these third party firms, they are
01:44:05
incentivized to pit workers against each
01:44:07
other because they want this data
01:44:10
annotation to happen at speed and as
01:44:12
cheaply as possible so that they can
01:44:14
also compete with one another in this
01:44:16
middle layer to get the the the bid the
01:44:19
the contract from the the client. And so
01:44:24
all of these workers that were
01:44:25
interviewed for this New York Magazine
01:44:27
story talk about how they actually no
01:44:29
longer have an ability to be human
01:44:32
because they are waiting at their laptop
01:44:35
to be pinged on Slack for when a project
01:44:38
is going to open up for data annotation
01:44:40
because they've tried job hunting. They
01:44:42
literally can't find anything else. This
01:44:44
is the thing that's going to help them
01:44:45
put food on the table for their kids.
01:44:46
And there was this one woman who said
01:44:49
like, "I have so much anxiety about when
01:44:52
the project is going to come, when it's
01:44:54
going to leave that when the project
01:44:56
came, it was right when my kid was
01:44:58
coming off of off of school." And I just
01:45:01
started tasking furiously because I
01:45:03
don't know what's going to go and I need
01:45:04
to earn as much money as possible in
01:45:05
this window of opportunity. So then my
01:45:07
when my kid came home and tried to talk
01:45:10
to me, I screamed at my child for for
01:45:13
distracting me. And then she was like,
01:45:16
"I've become a monster and I'm not even
01:45:19
allowed to go to the bathroom or take
01:45:22
care of my kids, let alone myself,
01:45:25
because this industry that is absorbing
01:45:28
more and more of the workers that are
01:45:30
being laid off, is mechanizing my life,
01:45:34
atomizing my work, devaluing my
01:45:38
expertise, and then harvesting it for
01:45:42
the perpetuation of this machine that
01:45:44
all of these AI executives are saying is
01:45:46
then going to come for everyone else's
01:45:48
jobs. And so what you were saying about
01:45:52
these this class of workers,
01:45:54
the business owners that get to become
01:45:57
more human because there are all of
01:45:59
these AI models now doing the tasks that
01:46:01
they don't have to do anymore. It is at
01:46:03
the cost of the vast majority of people
01:46:06
who are not business owners that are
01:46:09
struggling to find work getting absorbed
01:46:11
into the work of then providing these
01:46:15
technologies that the business owners
01:46:16
can use
01:46:18
>> and instead of becoming more human they
01:46:21
feel like their humanity has been
01:46:23
squeezed and diminished and they have no
01:46:27
ability to have control, agency and
01:46:30
dignity in their lives anymore. I think
01:46:32
this is a big I think this is a big
01:46:33
question that kind of pertains to this
01:46:34
graph here which is you know all of
01:46:37
these people if we believe anthropics
01:46:39
prediction of who will be disrupted
01:46:41
these people in these industries like
01:46:43
arts and media legal um life and social
01:46:47
sciences architecture and engineering
01:46:49
computer and maths business and finance
01:46:52
and management and also office and
01:46:54
admin. These people if we believe this
01:46:56
would have to retrain at something else
01:46:58
and unlike the industrial revolution
01:47:00
where you might get 10 20 years to
01:47:01
retrain because factories take a long
01:47:03
time to build. The distribution layer
01:47:05
that AI sits on top of is the open
01:47:06
internet. So this is why chat can go and
01:47:09
get hundreds of millions of users in no
01:47:11
time at all and become the fastest
01:47:12
growing company of all time. Um one of
01:47:15
my fears is that this disruption takes
01:47:17
place at a speed where we can't
01:47:20
transition.
01:47:21
And that was you know that I think you
01:47:23
you you said that sentence in the
01:47:25
passive voice the transition would
01:47:28
happen at a speed but who is driving
01:47:31
that speed?
01:47:32
>> Um
01:47:33
>> it's the companies
01:47:34
>> and their race with one another.
01:47:36
>> Yeah. And so they are driving the
01:47:38
transition to happen at a speed at which
01:47:42
it would be really hard to take care of
01:47:46
all of the people that would be
01:47:47
bulldozed over by
01:47:49
>> this is one of the crazy questions that
01:47:50
no one can answer for me when I sit with
01:47:52
these people that are AI CEOs. So I go,
01:47:54
"So what happens to the people if this
01:47:55
is if you agree that this is going to
01:47:56
happen at super speed?" You know, I
01:47:58
spoke to that CEO of Uber, Dar, who said
01:48:00
very similar things to what you're
01:48:01
saying is, you know, there'll be data
01:48:03
labeling jobs, for example, for the
01:48:04
drivers. But um they can't all become
01:48:07
data labelers. And there's a question
01:48:09
around meaning and purpose and
01:48:10
fulfillment. And that comes from losing
01:48:13
your meaning in life. I s also sit here
01:48:15
with so many people who talk about how
01:48:17
their father lost their job in Iran or
01:48:19
some some other country and came to the
01:48:22
United States and had to be a a toilet
01:48:24
cleaner on particular case was a doctor
01:48:26
in Iran but came to the US and was a
01:48:28
toilet cleaner and had to deal with the
01:48:30
sense of shame that that particular
01:48:31
person felt and the lack of dignity that
01:48:33
that caused and how that made that
01:48:35
person's self-esteem feel and the
01:48:36
depression alcoholism that transpired
01:48:38
from that. um if this happens at a large
01:48:40
scale across society, there's going to
01:48:43
be a ton of consequences like that.
01:48:45
>> I mean, this is this is like the core
01:48:47
themes of my work. And the reason why
01:48:49
I'm critical of these companies is that
01:48:50
they are creating technologies in a way
01:48:53
that creates the halves and have nots in
01:48:56
an extreme form that we have. It's it's
01:48:59
exacerbating the inequality that we
01:49:01
already see in the world. Like the
01:49:03
people who have things will have way
01:49:07
more riches. they'll have way more free
01:49:08
time. They'll be allowed to be more
01:49:10
human. But the people who don't have
01:49:12
things are even being squeezed even
01:49:16
more. And it's not just from a work
01:49:20
perspective. I mean, I talk in my book
01:49:23
also about the environmental and public
01:49:25
health crisis that these companies have
01:49:27
created where they are building these
01:49:31
colossal supercomput facilities. there
01:49:35
and and in in comm community like
01:49:37
communities all around the world and
01:49:39
they specifically pick some of the most
01:49:41
vulnerable communities. We're sitting in
01:49:42
Texas right now. Open AAI's largest one
01:49:46
of its largest data center projects is
01:49:48
being built in Abalene, Texas as part of
01:49:50
the Stargate initiative which was an
01:49:52
effort announced at the beginning of
01:49:54
Trump's second administration to spend
01:49:56
$500 billion on AI computing
01:49:58
infrastructure.
01:50:00
This facility
01:50:02
consumes will when it's finished will
01:50:05
consume more than a gigawatt of power
01:50:07
which is over 20%
01:50:11
over 20%. So this is actually a little
01:50:13
bit inaccurate now. Um this was
01:50:15
something that circulated online for a
01:50:17
while but there's updated numbers
01:50:18
>> just for someone that can't see cuz
01:50:20
they're listening on Spotify or
01:50:21
something. It's a picture of the size of
01:50:23
this facility.
01:50:25
>> So this is not the Abene Texas one. This
01:50:28
is a meta facility. Yeah. So, let's
01:50:29
first talk about opening eyes facility
01:50:31
in Texas. That one would be the size of
01:50:34
Central Park and it would run a million
01:50:37
computer chips and it would require the
01:50:41
power of more than 20% of New York City.
01:50:45
>> Do you know one of the things which I
01:50:47
found confusing, so I'd like to like
01:50:48
alleviate the dissonance is I thought
01:50:50
you were saying earlier that you didn't
01:50:51
think the job disruption promises were
01:50:53
real.
01:50:55
No, what I was saying is that when we
01:50:59
talk about what these executives predict
01:51:03
about the future, we need to understand
01:51:05
that they are ultimately trying to
01:51:08
influence the public in a way that
01:51:10
allows them to continue maintaining
01:51:11
control over the technology.
01:51:13
>> But objectively, do you think that the
01:51:14
job disruption that they talk about
01:51:16
where
01:51:16
>> Yeah. Yeah. I mean I I mentioned
01:51:18
>> real
01:51:18
>> well I
01:51:20
>> I don't want to comment specifically on
01:51:21
like this chart but it's like we've
01:51:23
already seen in job reports that there
01:51:25
is a restructuring of the economy
01:51:27
happening right now. Yeah.
01:51:28
>> But but going back to like the data
01:51:30
center. So this supercomputer facility
01:51:32
it's a meta supercomputer facility
01:51:34
>> is being built in Louisiana
01:51:37
>> and it would be four times the size of
01:51:39
the Abene Texas one and use half of the
01:51:43
average power demand of New York City.
01:51:44
So it's one the size of Manhattan. This
01:51:46
makes it seem like almost all of
01:51:48
Manhattan, but it's it would be 1/5 the
01:51:49
size of Manhattan. When these facilities
01:51:52
go into these communities, what happens?
01:51:55
Power utility increases, grid
01:51:57
reliability decreases. The facilities
01:52:01
also need fresh water to generate the
01:52:04
power for powering them as well as fresh
01:52:06
water to cool. And there have been lots
01:52:08
of documented stories of communities
01:52:10
that are already really constrained in
01:52:12
their freshwater resource. they're under
01:52:13
a drought when a facility comes in and
01:52:15
then there are people the community is
01:52:17
actually like competing with this
01:52:19
facility for fresh water. I talk about
01:52:20
one of those communities in my book and
01:52:22
also sometimes these facilities instead
01:52:25
of connecting to the grid they instead a
01:52:29
a power plant pops up next to it. So in
01:52:31
Memphis Tennessee where Musk built
01:52:34
Colossus the supercomputer for training
01:52:36
Grock he used 35 methane gas turbines to
01:52:41
power the facility. This is a
01:52:42
working-class community, a black and
01:52:44
brown community, a rural community that
01:52:47
was not even told that they would be the
01:52:49
hosts of this facility. And they
01:52:52
discovered it because they literally
01:52:54
smelled what seemed like a gas leak in
01:52:58
all of their living rooms. And that's
01:52:59
when they discovered that these methane
01:53:02
gas turbines were taking away their
01:53:05
right to clean air. And this is a
01:53:08
community that's already been facing a
01:53:10
history of environmental racism. They
01:53:12
had already had lots of struggles to
01:53:15
access their right to clean air. And now
01:53:18
there's this huge supercomput that's
01:53:21
landed in their midst that is pumping
01:53:24
thousands of tons of toxins into their
01:53:27
air, exacerbating the asthmatic symptoms
01:53:30
of the children, exacerbating the
01:53:32
respiratory illnesses of other people.
01:53:35
that it's it's one of the communities
01:53:36
that has the highest rates of um lung
01:53:40
cancer
01:53:41
and so
01:53:42
>> and that supercomputers taking their
01:53:44
jobs
01:53:45
>> and then they also have supercomputers
01:53:46
taking their jobs. So, so this is what I
01:53:48
mean is like the halves and have nots
01:53:51
are fundamentally
01:53:53
being pulled apart even further. Like if
01:53:56
you in this version of Silicon Valley's
01:53:59
future are in the misfortunate category
01:54:03
of being a have not, we are talking
01:54:06
about you now getting a job that is way
01:54:09
worse than what you had because you
01:54:11
might be doing data annotation
01:54:13
>> and you might be treated as a machine
01:54:16
rather than as a human to extract value
01:54:18
the value of your labor for perpetuating
01:54:20
this labor automating machine that these
01:54:23
people are building. You might be
01:54:26
competing with these facilities for
01:54:28
freshwater resources. They're also
01:54:30
polluting your air. Your bills have
01:54:32
increased. So, the affordability crisis
01:54:34
is getting worse.
01:54:37
Like, how is that making people able to
01:54:40
be more human?
01:54:41
>> What do we do about it?
01:54:43
>> Yes.
01:54:45
>> Okay. So, one of the analogies that I
01:54:47
always use is AI is like the word
01:54:50
transportation. Transportation can
01:54:52
literally refer to everything from a
01:54:53
bicycle to a rocket. And we have nuanced
01:54:57
conversations about transportation where
01:54:59
we always say we need to transition our
01:55:01
transportation towards more uh
01:55:05
sustainable options. We need a
01:55:06
transition towards you know public
01:55:08
transport, electric vehicles. And we
01:55:11
don't we don't ever say everyone should
01:55:13
get a rocket to do every to serve all of
01:55:16
their transportation needs, right? Like
01:55:18
we're in Austin. If you use a rocket to
01:55:20
fly from Dallas to Austin, like that
01:55:22
would just make not no sense. It's just
01:55:24
a disproportionate use of resources to
01:55:26
get the benefit
01:55:28
of getting from point A to point B. This
01:55:31
how we should think about AI. So all of
01:55:33
the models that we've been talking
01:55:35
about, I like to think of them as the
01:55:37
rockets of AI. They use an extraordinary
01:55:40
amount of resources and they provide
01:55:41
benefit some dramatic benefit to some
01:55:44
people but they're also exacting an
01:55:47
extraordinary cost on a large swath of
01:55:49
people because of the like the costs of
01:55:53
developing this technology.
01:55:57
Why don't we build more bicycles of AI?
01:56:00
This is things like deep minds alpha
01:56:02
fold which is a system that predicts how
01:56:06
proteins will fold based on amino acid
01:56:08
sequences. It's really important for
01:56:10
accelerating drug discovery for
01:56:14
understanding human disease and it won
01:56:15
the Nobel Prize in chemistry in 2024.
01:56:18
And the reason why it's a bicycle of AI
01:56:20
is because you're using small curated
01:56:23
data sets. you're just you just have
01:56:26
data that has amino acid sequences and
01:56:29
protein folding. So that means you need
01:56:32
significantly less computational
01:56:35
resources to develop the system, which
01:56:36
means significantly less energy, which
01:56:38
means less emissions, so on and so
01:56:39
forth. And you're providing enormous
01:56:42
benefit to people.
01:56:43
>> It feels like the
01:56:46
horse has left the stable in this regard
01:56:48
because they've already taken people's
01:56:50
IP, they've taken media, they they train
01:56:52
on this podcast. We know they do because
01:56:54
it it shows that they do. Um I think
01:56:56
there's a button actually in the back
01:56:57
end of YouTube now that allows you just
01:56:58
to click it and it says we will train on
01:57:00
your YouTube channel. Um so the horses
01:57:04
kind of left.
01:57:04
>> Here's the thing. If the horse truly had
01:57:06
left the stables, they wouldn't have to
01:57:08
train on anything anymore. Why is it
01:57:10
that their appetite for data has
01:57:12
actually expanded? It's because in order
01:57:15
to build the next generations of their
01:57:17
technologies, in order to have the
01:57:18
technologies continue to be relevant and
01:57:21
continue to update with the pace of new
01:57:25
knowledge creation and society's
01:57:27
evolvement, they need to train again and
01:57:30
again and again and again. And why are
01:57:33
they employing actually more and more
01:57:35
and more data annotation workers over
01:57:36
time? It's because they need more and
01:57:39
more of that work over time. I mean,
01:57:41
I've been reporting on data annotation
01:57:44
work for over 7 years now, and it's not
01:57:47
gone down. It's gone it's increased.
01:57:50
>> Do you think there's any chance of it
01:57:52
going down? Do you think there's any
01:57:54
chance of this sort of brute force
01:57:55
scaling approach where you take data,
01:57:57
you take computational power, energy,
01:58:00
and you, you know, you have um the data
01:58:04
labelers and, you know, building out
01:58:05
more and more parameters for the models.
01:58:07
Do you think there's any chance it's
01:58:09
going to stop or go in a different
01:58:10
direction other than the one it's going
01:58:11
in now?
01:58:12
>> I would love to reframe the question and
01:58:14
say what should we be doing in this
01:58:16
moment where it's not going down where
01:58:19
we do recognize that actually these
01:58:21
companies in this moment need continued
01:58:24
resources, inputs and labor to
01:58:26
perpetuate what they are doing.
01:58:28
>> Yeah. because this sounds like stop
01:58:30
>> and I just feel like stop is like a HUD.
01:58:33
It feels like I just think you know with
01:58:35
the government in place they're
01:58:36
supporting these companies like crazy.
01:58:37
Globally this is happening. So I'm like
01:58:40
stop doesn't feel
01:58:41
>> I always say we need to break up the
01:58:43
empire and we need to develop
01:58:44
alternatives and we are already seeing a
01:58:47
flourishing of incredible grassroots
01:58:50
movements that are applying an enormous
01:58:52
amount of pressure to the way that the
01:58:54
empire is trying to unfold its agenda.
01:58:58
80% of Americans in the most recent poll
01:59:00
think that the AI industry need to be
01:59:02
regulated.
01:59:03
>> Yeah.
01:59:04
>> When was the last time that 80% of
01:59:05
Americans were on the same side of an
01:59:07
issue?
01:59:07
>> No. Yeah. When I have these
01:59:08
conversations on the podcast, the
01:59:09
comment section are clear.
01:59:10
>> Yeah.
01:59:11
>> There's no there's no disagreement.
01:59:12
There's no one in there going, "Oh, no.
01:59:13
I think they should crack on."
01:59:14
>> Yeah. Dozens dozens of protests against
01:59:17
data centers have broken out all around
01:59:19
this country and the US, all around the
01:59:21
world.
01:59:22
>> So, what do we do about it?
01:59:23
>> So, these are thing people that are
01:59:25
doing something about it. They are
01:59:27
actually reasserting their agency and
01:59:30
exercising democratic contestation
01:59:33
against the ways that the empires are
01:59:35
going about their business.
01:59:36
>> What goal should we be aiming at? So, if
01:59:38
I said to my audience, Janet at home,
01:59:40
because this is kind of what I see in
01:59:41
the comments, it's hopelessness. It's
01:59:42
like, what can I do? I'm just a
01:59:44
>> Yeah. Well, well, well, the goal is not
01:59:47
that we completely get rid of this
01:59:49
technology. The goal is that these
01:59:50
companies need to stop being empires.
01:59:52
And the way I define like a typical
01:59:53
business versus an empire is that the
01:59:55
empires are predicated on this idea that
01:59:58
they do not have to provide a fair
02:00:00
exchange of value with the workers who
02:00:02
work for them or the people who use them
02:00:04
or all of the other people that are
02:00:05
involved in like the supply chain of
02:00:07
producing and deploying these
02:00:08
technologies. They can extract and
02:00:10
exploit and extract and exploit and get
02:00:12
more value than what they offer. Whereas
02:00:15
typical businesses, there's a fair
02:00:16
exchange. you you buy a service, you
02:00:19
feel like you got the same amount of
02:00:20
value as the service that you provided.
02:00:22
But like for these data annotation
02:00:23
workers, for example, they do not feel
02:00:25
in any way that they're being paid the
02:00:27
same value that they provide to these
02:00:28
companies. So that's like for me the
02:00:30
north star is like we should be pushing
02:00:33
back and holding accountable these
02:00:36
companies when they operate in an
02:00:38
imperial way. And that's what we've seen
02:00:41
with all of these people that are now
02:00:43
literally protesting in the streets
02:00:44
against data centers and having an
02:00:46
enormous effect, by the way, actually
02:00:48
stalling data center projects and also
02:00:51
completely banning data centers from
02:00:53
being developed in their localities.
02:00:54
We're seeing that with artisan writers
02:00:56
that are suing these companies for
02:00:59
intellectual property infringement and
02:01:00
creating a huge public conversation
02:01:02
about what is it that we actually how do
02:01:05
we actually want to protect our
02:01:06
intellectual property? It's like I three
02:01:09
weeks ago I met Megan Garcia who is the
02:01:12
mother of Sul Settzer III who is the
02:01:16
14-year-old who died by suicide after
02:01:19
being sexually groomed by a
02:01:21
characterized chatbot.
02:01:23
And she when that happened
02:01:27
I mean obviously was incredibly
02:01:30
devastated by what had happened to her
02:01:32
son. She also decided to do something
02:01:35
about it. She sued the companies and
02:01:37
that lawsuit then sparked many other
02:01:39
parents and families who were actually
02:01:41
experiencing similar things to sue these
02:01:44
companies as well. That has created an
02:01:46
enormous public conversation about what
02:01:50
these companies are actually doing when
02:01:52
they exploit and they extract. What is
02:01:55
the cost to the lives of people around
02:01:58
the world including children? So, what
02:02:01
do you think my audience should do if
02:02:02
they if they agree with everything
02:02:03
written in your book, Age Empire of AI,
02:02:06
Dreams and Nightmares, and Sam Mortman's
02:02:08
Open AI? If they agree with everything
02:02:10
said here, if they agree with everything
02:02:11
we've discussed today, they're concerned
02:02:13
about their kids, they they don't want
02:02:15
everyone to become data labelers, they
02:02:17
don't think that's a, you know,
02:02:18
particularly great solution, what what
02:02:20
can they actually go and do?
02:02:22
>> When I was writing the book, the only
02:02:24
discourse that was happening was this is
02:02:26
the best thing since sliced bread.
02:02:27
>> Mhm. because of all of the actions of
02:02:30
these people like saying when they're
02:02:32
comp they're they're not happy with the
02:02:35
things that these companies are doing.
02:02:37
We now have 80% of Americans that want
02:02:39
to regulate this industry. And so I
02:02:40
would say to people, think about all of
02:02:43
the ways that your life intersects with
02:02:46
the resources and the that the AI
02:02:49
industry needs to perpetuate what they
02:02:50
do and also the spaces that they would
02:02:53
need to deploy these technologies to
02:02:55
continue having broad-based adoption
02:02:58
>> in their work. So you're a data donor to
02:03:02
these companies. You could withhold that
02:03:05
data. And that's what those artists and
02:03:07
writers are are doing. like they're
02:03:08
suing these companies to withhold to try
02:03:10
and create mechanisms by which that data
02:03:12
would then be withheld. You probably
02:03:15
have a data center popping up around
02:03:16
you. If you're at a school environment
02:03:19
or a company environment, you're
02:03:21
probably having a discussion in those
02:03:23
environments right now about what should
02:03:25
the AI adoption policy be? And these
02:03:27
companies they like I was talking with
02:03:30
some open air employees just the other
02:03:32
day and they were telling me that it's
02:03:35
understood internally that the revenue
02:03:38
targets for the company are
02:03:41
extraordinary and they need things to go
02:03:44
flawlessly for it to all work out. And
02:03:48
so they would need every single person
02:03:51
to adopt this, every single space to
02:03:53
adopt this. They would need to be able
02:03:55
to build their data centers at the speed
02:03:57
that they're trying to build them. And
02:03:59
so what I would say to everyone of your
02:04:01
viewers is let's not make it go
02:04:03
flawlessly if we don't agree with what
02:04:05
they are doing.
02:04:06
>> Ah, okay. I got you.
02:04:08
>> And then let's build alternatives.
02:04:09
Because
02:04:11
the thing is what I'm saying is not that
02:04:14
these technologies don't have utility.
02:04:16
It's that specifically the political
02:04:18
economy that has emerged to support the
02:04:20
production of these technologies right
02:04:22
now
02:04:23
>> is exacting a lot of harm on people. But
02:04:25
we have research that shows that the
02:04:28
very same capabilities could be
02:04:31
developed with much more efficient
02:04:33
methods with much less resource
02:04:35
consumption. And we have a lot of
02:04:38
different other AI systems at our
02:04:40
disposal that are like the bicycles of
02:04:41
AI that we also know provide
02:04:44
extraordinary benefit at very little
02:04:46
cost. So let's break up the empire and
02:04:48
let's forge new paths of AI development
02:04:50
that are broadly beneficial to everyone.
02:04:53
>> It's strange. I'm quite I think I'm I'm
02:04:56
I've trained myself to deal with
02:04:58
dichotoies in my head. And this for me
02:05:00
is such is a dichotomy where I as a CEO
02:05:04
and as a founder, as an entrepreneur and
02:05:05
someone that loves technology, I think
02:05:07
it's incredible. It's absolutely
02:05:08
incredible AI. It's just so amazing and
02:05:11
incredible the things it's enabled me to
02:05:12
do and create.
02:05:13
>> Yeah. Because it's designed to enable
02:05:15
people like you.
02:05:16
>> And my car driving in the morning and
02:05:19
being safer. Incredible. Um I think you
02:05:23
know the billion odd people that use AI
02:05:25
tools or chat or whatever it might be,
02:05:26
they'd probably say that it's added
02:05:28
value to their life. But and this is the
02:05:30
part that people find confusing that you
02:05:32
can and I like I invest in companies
02:05:33
that are you know heavily using AI but
02:05:36
and the big butt is is it possible to
02:05:37
think that is true and also think that
02:05:40
there are significant unintended
02:05:42
consequences which technology in the
02:05:44
history of technology should have taught
02:05:45
us to take a moment to pause to talk
02:05:47
about because
02:05:48
>> I think this is absolutely like you can
02:05:52
have both of these things in your head
02:05:53
and what I'm saying is that this tension
02:05:55
doesn't have to be a tension because we
02:05:58
could actually preserve the utility and
02:06:01
benefits of these technologies but
02:06:03
actually develop and design them in a
02:06:05
different way that doesn't have all of
02:06:07
these unintended consequences.
02:06:09
>> Yes. And I think there needs to be a big
02:06:10
social conversation which is why I have
02:06:12
so many conversations about AI in the
02:06:13
show like there needs to be a big social
02:06:14
conse uh conversation about being
02:06:17
intentional about the social impact um
02:06:20
the social and environmental impact and
02:06:22
that conversation is not being had in
02:06:23
the in government. From what I can see,
02:06:26
the conversation takes place in the
02:06:28
industry and actually trying to pull it
02:06:30
out of the industry and and open
02:06:31
people's minds to it is hopefully what
02:06:33
we've been doing over the last couple of
02:06:34
months with this subject because
02:06:35
>> I think it's actually been it it has
02:06:38
been been happening everywhere outside
02:06:40
of the industry and for local
02:06:42
governments and state level governments
02:06:44
there have been huge conversations about
02:06:46
this everywhere. Like I've been on book
02:06:48
tour, I've been to dozens of cities
02:06:49
around the world. People are having
02:06:53
these crucial conversations everywhere.
02:06:56
I have not gone to a single city.
02:06:57
>> Yes. Everywhere. Even here in South by.
02:06:59
>> Yeah. I haven't gone to a single city
02:07:00
where the room is not packed and people
02:07:03
are not wrestling with the same exact
02:07:04
questions as every other person in every
02:07:06
other room that I've been in.
02:07:08
>> Speaking of packed rooms, I know you've
02:07:09
got to go cuz you've got you've got to
02:07:11
talk today. So, I'm going to we've got a
02:07:13
last question which is the closing
02:07:14
tradition on this podcast. How would
02:07:15
your advice to a friend with a terminal
02:07:17
diagnosis differ from what you would do
02:07:21
yourself?
02:07:22
>> That's a great question.
02:07:24
>> Differ from what you would do yourself?
02:07:25
>> Oh my god. I have
02:07:28
I I would tell them like enjoy
02:07:31
like live life for yourself. Um you
02:07:33
wouldn't do it
02:07:34
>> and take it easy. And yeah, I I I
02:07:38
am not taking it easy.
02:07:39
>> Well, I think it's a good thing you're
02:07:40
not taking it easy because you're
02:07:41
leading a conversation which is
02:07:42
incredibly important. And I think that's
02:07:44
the thing. I think the conversation is
02:07:46
the important thing. And so, you know,
02:07:49
because of algorithms and echo chambers,
02:07:50
it's so rare to have a conversation
02:07:52
>> these days, especially a long form one.
02:07:54
I agree.
02:07:55
>> Like this. So, I think they're so
02:07:56
important. And your book is for anyone
02:07:58
that's curious about
02:07:59
>> I think a lot of people would have
02:08:01
learned a lot of stuff today cuz I sit
02:08:03
here with and interview AI people all
02:08:04
the time and I've learned so much today.
02:08:06
From reading your book and the extensive
02:08:08
objective perspective that your book
02:08:10
takes, you you're able to unravel all of
02:08:12
these stories that we sometimes see in
02:08:14
tweets and we don't know if they're true
02:08:15
or not because you've gone and met the
02:08:16
people and you've done your research and
02:08:18
you're incredibly intelligent person,
02:08:20
extremely intelligent person who clearly
02:08:23
has humanity's interests as your north
02:08:26
star and that shows up in everything you
02:08:28
do and everything you say. So please
02:08:29
continue to fight in the way that you
02:08:30
are um because it's an incredibly
02:08:32
important one. people like you that are,
02:08:35
I think,
02:08:36
galvanizing the world to take the
02:08:39
collective action that we're starting to
02:08:40
see everywhere.
02:08:42
>> Yeah.
02:08:42
>> Empire of AI: Dreams and Nightmares in
02:08:44
Sam Alman's Open AI by Karen How. I'll
02:08:47
link it below for anyone that wants to
02:08:49
read this book. I highly recommend you
02:08:50
do. It's a New York Times bestseller for
02:08:51
good reason. Karen, thank you.
02:08:53
>> Thank you so much, Stephen.
02:08:54
>> YouTube have this new crazy algorithm
02:08:56
where they know exactly what video you
02:08:58
would like to watch next based on AI and
02:09:01
all of your viewing behavior. And the
02:09:02
algorithm says that this video is the
02:09:06
perfect video for you. It's different
02:09:07
for everybody looking right now.
