00:00:00
We should talk about this this um AI privacy issue that Sam Alman mentioned. All right, that's a great segue because
00:00:05
I saw that as well, David Saxs, and uh as our civil servant working on AI, this is something where you can have an
00:00:11
additional contribution. There's more work we can give you. All right, listen. Here it is. AI user privacy is becoming
00:00:17
an issue because friend of the pod Sam Alman says there is no legal confidentiality when using his product
00:00:25
Chat GPT. Here's a 30 second clip again. Friend of the pod fop Sam Alman on Theo
00:00:33
Vaughn. People talk about the most personal in their lives to chatt young people
00:00:37
especially like use it as a therapist, a life coach, uh having these relationship
00:00:41
problems. What should I do? And right now if you talk to a therapist or a lawyer or a doctor about those problems,
00:00:48
there's like legal privilege for it. We haven't figured that out yet for when you talk to Chad GBT. So if you go talk
00:00:53
to CHP about your most sensitive stuff and then there's like a lawsuit or whatever, like we could be required to
00:00:58
produce that and I think that's very screwed up. I think we should have like the same concept of privacy for your
00:01:03
conversations with AI that we do with a therapist or whatever. Okay, Saxs, this is bringing up
00:01:09
something super important. What's your take on it? Okay. Well, I I think this is an interesting topic
00:01:14
because like copyright, this is an area where we have existing law, but it does make you rethink whether those laws are
00:01:23
truly applicable or make as much sense in this new world. So, the existing law, the existing example is search history.
00:01:32
You know, the government can get a copy of your search history. They can subpoena it.
00:01:36
Yeah. Every true crime story starts with the person's search for how do I kill my
00:01:41
husband slowly with you know poison and then they Yeah. That's right. Exactly. The point is though that
00:01:48
I think Sam is right about the legal treatment right now which is that your chat history isn't any different than
00:01:55
the search history in the eyes of the law. But it is much more personal. It's much more interactive than your search
00:02:01
history. You are using it like you said. You could use it as a as your doctor. You could use it as your therapist. You
00:02:07
could use as your lawyer. And so the ability for the federal government to be intrusive is so much greater than with
00:02:15
your search history. So I don't know what like the right policy should be yet, but I I will say it does make me
00:02:23
uncomfortable. Yeah, there's a market. Can I make a recommendation to my AIR? Yes, please. He's our
00:02:30
Why don't we let AI models get bar certified and get medically certified? So if the AI models, it turns out, are
00:02:40
actually proving to be more accurate, more thoughtful, more responsive, more reasonable,
00:02:47
whatever it is, whatever metric we're using, and they pass the same criteria as one would need to pass to qualify for
00:02:53
the bar or to qualify for a doctor certificate. Why don't we do that for the AI? If that then happens, then the
00:03:00
the same privilege occurs to the AI as it does to the individual human that does it. And now if you extrapolate from
00:03:05
where that takes us, if we're suddenly giving AI the same sort of privileged rights that we give to privileged
00:03:10
humans, where's that going to take us ultimately with respect to the overall rights for AI?
00:03:14
Well, and they have responsibility. Hold on a second. I just want to point out here once again, you have a mind-blowing
00:03:20
concept here. I've never heard anybody vocalize that. Could they actually be certified in that knowledge? And if they
00:03:26
pass the test, makes sense they would. But then you also get responsibility. So with great power comes great
00:03:30
responsibility. I will tell you this. You can turn this stuff off. But this is an opportunity. I'm going to send a note
00:03:35
to El. It sounds crazy today, but I guarantee if you put it on Poly Market, there will
00:03:39
be a date when this happens. Let's do it. Poly market. Shout out to Shane. Let's get that up there. I just
00:03:43
want to point out I'm going to email Elon about this when I get off the pod. This is an opportunity to create the
00:03:47
signal of the signal equivalent of an LLM. The all of your chat should be encrypted. All of it should be by by
00:03:55
default. encrypt it by default on Grock. Make it so that Grock can't even see it.
00:04:00
They don't have it. So when you try to subpoena it, you can do what Tim Cook does, which he says like I don't have
00:04:04
it. You if you want to try to back door it, you can. That's a market opportunity. I can tell you I only use
00:04:10
the Brave browser and Brave search for this reason. I don't want my search history like saved somewhere or
00:04:16
whatever. That you can take control of this as an individual, but the defaults matter and you have to then do the work.
00:04:22
It's a great market opportunity. Hey Chimoth, I don't even want to know what you're talking to chat GPT about. What
00:04:26
are you What's in your chat GBT logs? What's in there, Chimoth? How to extend? How to get the extra centimeter? What's
00:04:32
in there? You trying to What's in there? I keep asking it to find me a moderator.
00:04:38
Oh, great. I keep asking it to find me a participant who's not a douche. Um, my god, you are so deep in your villain
00:04:49
era and you're leaning into it and I'm so here for it. I love your villain. You know why I am so
00:04:55
Why are you going into your villain era? I am so riskon right now. It's like it's liberating actually. It's amazing.
00:05:03
It's really amazing. Is there any blowback to how outlandish you've become this year? Any blowback at
00:05:09
all? Has it had any negative consequence on business or hiring or anything? No. But but outlandish? How how have I
00:05:15
been outlandish? You're you're you're just filter off. You're filter off as And I think it's
00:05:20
great. I think the over two windows back. It's absolutely fantastic we're seeing here. I asked Chad GPT about my
00:05:25
future and uh my IQ. It's very interesting when you ask ChatGpt to analyze you. I suggest everyone do it.
00:05:30
Well, actually, yeah, when you just ask ChatGpt or whatever, what do you know about me?
00:05:36
And it's scary how much it already does. It's scary. There's this great personality test. You can put this
00:05:40
personality test into Grock and this guy like made this prompt and it goes and it
00:05:45
tells you all your personality based on your Twitter ex history. It is wild how accurate it is. What does it say about
00:05:51
you, Jal? I'm actually curious. It says the same thing about all of us. We're all likeworked
00:05:57
narcissist ENTJ. You can literally run the MyersBriggs against your Yeah. your chat history. It's actually But I I
00:06:05
like your mind-blowing concept there, by the way, of like them becoming certified
00:06:09
in some