E124: AutoGPT's massive potential and risk, AI regulation, Bob Lee/SF update

1. 0:00 – 1:49

   Bestie intros!

   1. JCJason Calacanis0:00

      Welcome to episode 124 of the All In podcast. My understanding is there's gonna be a bunch of global fan meetups for episode 125. If you go to Twitter and you search for All In fan meetups, you might be able to find the link.

   2. DSDavid Sacks0:14

      But just to be clear, we're not, they're not official All In. It's-

   3. JCJason Calacanis0:17

      They're not.

   4. DSDavid Sacks0:18

      ... they're fans. It's self-organized, which is pretty mind-blowing. But we can't vouch for any particular organization, right?

   5. JCJason Calacanis0:25

      Nobody knows what's gonna happen at these things.

   6. DSDavid Sacks0:26

      (laughs)

   7. JCJason Calacanis0:26

      You could get robbed. (laughs) It could be a setup. I don't know.

   8. DSDavid Sacks0:29

      (laughs)

   9. JCJason Calacanis0:30

      But I retweeted it anyway because there are-

   10. DSDavid Sacks0:32

       (laughs)

   11. JCJason Calacanis0:32

       ... 31 cities where you lunatics are getting together to celebrate the world's number one business technology podcast.

   12. DSDavid Sacks0:40

       It is pretty crazy. You know what this reminds me of, is in the early '90s when Rush Limbaugh became a phenomenon-

   13. JCJason Calacanis0:46

       Uh-huh.

   14. DSDavid Sacks0:47

       ... there used to be these things called Rush Rooms where, like, restaurants and bars would literally broadcast Rush over their speakers during, I don't know, like, what, for the morning through lunch broadcast. And people would go to these Rush Rooms and listen together.

   15. JCJason Calacanis1:01

       What was it like, Sax, when you were about 16, 17 years old at the time? (laughs)

   16. DSDavid Sacks1:04

       (laughs)

   17. JCJason Calacanis1:04

       What was it like when you hosted this?

   18. DSDavid Sacks1:05

       It was a phenomenon. But I mean, it's-

   19. JCJason Calacanis1:07

       Really?

   20. DSDavid Sacks1:07

       ... kind of crazy. We've got, like, a phenomenon going here where people are-

   21. JCJason Calacanis1:10

       I love it. You ju-

   22. DSDavid Sacks1:10

       ... self-organizing.

   23. CPChamath Palihapitiya1:12

       You've said phenomenon three times-

   24. DSDavid Sacks1:14

       (laughs)

   25. CPChamath Palihapitiya1:14

       ... instead of phenomenon.

   26. DSDavid Sacks1:15

       (laughs)

   27. CPChamath Palihapitiya1:15

       He said it's phenomenon.

   28. JCJason Calacanis1:17

       Phenomenal. Why is Sax in a good mood, Shema? What's going on?

   29. CPChamath Palihapitiya1:20

       There's a specific secret toe tap that you do under the bathroom stalls when you go to a Rush Room.

   30. DSDavid Sacks1:25

       (laughs)
2. 1:49 – 23:57

   Understanding AutoGPTs

   1. JCJason Calacanis1:49

      There's a lot of actual news in the world and generative AI is taking over the dialogue and it's moving at a pace that none of us have ever seen in the technology industry. I think we'd all agree. The number of companies releasing product and the compounding effect of this technology is phenomenal, I think we would all agree. A product came out this week called Auto GPT and, uh, people are losing their mind over it. Basically what this does is it lets different GPTs talk to each other. And so you can have agents working in the background, and we've talked about this on previous podcasts, but they could be talking to each other essentially, and then completing tasks without much intervention. So if, let's say, you had a sales team and you said to the sales team, "Hey, look for leads that have these characteristics for our sales software, put them into our database, find out if they're already in the database, alert a salesperson to it, compose a message based on that person's profile on LinkedIn or Twitter or wherever, and then compose an email, send it to them. If they reply, offer them to do a demo and then put that demo on the calendar of the salesperson, thus eliminating a bunch of jobs." And you could run these, what would essentially be cron jobs in the background forever and they can interact with other LLMs in real time. Sax, I've just gave but one example here, but when you see this happening, give us your perspective on what this tipping point means.

   2. DSDavid Sacks3:24

      Let me take a shot at explaining it in a slightly different way.

   3. JCJason Calacanis3:28

      Sure.

   4. DSDavid Sacks3:28

      Not that your explanation was wrong, but I just think that maybe explain it in terms of something more tangible.

   5. JCJason Calacanis3:33

      Sure.

   6. DSDavid Sacks3:34

      So I, I had a friend who's a developer who's been playing with Auto GPT. By the way, so you can see it's on GitHub, it's kind of an open source project. It was sort of a hobby project, it looks like, that somebody put up there. It's been out for about two weeks. It's already got 45,000 stars on GitHub, which is a huge number.

   7. JCJason Calacanis3:51

      Explain what GitHub is for the audience.

   8. DSDavid Sacks3:53

      It's just a code repository and you can create, you know, repos of code for open source projects. That's where all the developers checking their code. So, you know, for open source projects like this, anyone can go see it and play with it.

   9. CPChamath Palihapitiya4:04

      It's like Pornhub, but for developers.

   10. DSDavid Sacks4:06

       (laughs)

   11. JCJason Calacanis4:07

       It would be more like amateur Pornhub because you're contributing your scenes, as it were, your code.

   12. DSDavid Sacks4:12

       Yes.

   13. JCJason Calacanis4:12

       But yes, continue.

   14. DSDavid Sacks4:12

       But in any event, this thing has a ton of, of stars and apparently, just last night, it got another 10,000 stars overnight. This thing is like exploding in terms of popularity. But in any event, what you do is you give it an assignment. And what Auto GPT can do that's different is it can string together prompts. So if you go to Chat GPT, you prompt it one at a time, and what the human does is you get your answer and then you think of your next prompt, and then you kind of go from there and you end up in a long conversation that gets you to where you want to go. So the question is, what if the AI could basically prompt itself? Then you've got the basis for autonomy, and that's what this project is designed to do. So, what you'll do is, what my friend did is he said, "Okay, you're an event planner AI, and what I would like you to do is plan a trip for me for a wine tasting in Healdsburg this weekend. And I want you to find, like, the best place I should go and it's got to be kid friendly. Not everyone's gonna drink. We're gonna have kids there. And I'd like to be able to have other people there, and so I'd like you to plan this for me." And so what Auto GPT did is, it broke that down into a task list, and every time it completed a task, it would add a new task to the bottom of that list. And so the output of this is that it searched a bunch of different wine tasting venues, it found a venue that had a bocce ball and lawn area for kids, it came up with a schedule, it created a budget, it created a checklist for an event planner. It did all these things and my friend says he's actually gonna book the venue this weekend and use it. So we're going beyond the ability just for a human to just prompt the AI, where now the AI can take on complicated tasks and, again, it can recursively update its task list based on what it learns from its own previous prompt. So what you're seeing now is the basis for-... a personal digital assistant. This is really where it's all headed, is that you can just tell the AI to do something for you pretty complicated and it will be able to do it. It will be able to create its own task list and get the job done, a- and quite complicated jobs. So that's why everyone's losing their shit over this.

   15. JCJason Calacanis6:23

       Freeberg, your thoughts on automating these tasks and having them run and, and add tasks to the list. This does seem like a sort of seminal moment in time that this is actually working.

   16. DFDavid Friedberg6:37

       I think we've been seeing seminal moments over the last couple of weeks and months kind of continuously. Every time we chat about stuff or every day, there's new releases that are paradigm-shifting and kind of reveal new applications and, and perhaps concepts structurally that we didn't really have a good grasp of before some demonstration came across. ChatGPT was kind of the, the seed of that and then all of this evolution since has really, I think, changed the landscape for really how we think about our interaction with the digital world and where the digital world can go and how it can interact with the physical world. It's, it's just really profound. One of the interesting aspects that I think I saw with some of the applications of AutoGPT were these almost, like, autonomous characters in, in, like, a game simulation that can interact with each other, or these autonomous characters that would speak back and forth to one another, where each instance, uh, has its own kind of predefined role and then it explores some set of discovery or application or prompt back and forth with the other agent. And that the kind of recursive outcomes with this agent-to-agent interaction model and perhaps multi-agent interaction model, again, reveals an entirely new paradigm for, you know, how things can be done simulation-wise, you know, discovery-wise, engagement-wise, where one agent, you know, each agent can be a different character in a room. And you can almost see how a team might resolve to create a new product collaboratively by telling each of those agents to have a different character background or different set of data or different set of experiences or different set of personality traits, and the evolution of those, that multi-agent system outputs, you know, something that's very novel, that perhaps any of the agents operating independently were not able to kind of reveal themselves. So again, like another kind of dimension of interaction with these, with these models. And it, uh, again, like every week, it's a whole nother layer to the onion. It's super exciting and compelling and the, the rate of change and the pace of kind of, you know, new paths being, being defined here really, I think, makes it difficult to catch up. And particularly, it highlights why it's gonna be so difficult, I think, for regulators to come in and try and set a s- set of standards and a set of rules at this stage, 'cause we don't even know what we have here yet.

   17. JCJason Calacanis9:04

       Mm-hmm.

   18. DFDavid Friedberg9:04

       And it's gonna be very hard to kind of put the genie back in the box.

   19. JCJason Calacanis9:08

       Yeah. And you're also referring, I think, to the Stanford and Google paper that was published this week. They did a research paper where they created essentially The Sims, if you remember that video game, put a bunch of, eh, what you might consider NPCs, non-playable characters, you know, the merchant or the whoever in a, um, in a video game, and they said, "Each of these agents should talk to each other, put them in a simulation." One of them decided to have a birthday party. They decided to invite other people, and then they have memories. And so then over time, they would generate responses like, "I can't go to your birthday party, but happy birthday." And then they would follow up with each player and seemingly emergent behaviors came out of this sort of simulation, which of course now has everybody thinking, well, of course we as humans, and this is simulation theory, are living in a simulation. We've all just been put into this. Chamath, is what we're experiencing right now how impressive this technology is? Or is it, oh wow, human cognition maybe we thought was incredibly special, but we can actually simulate a significant portion of what we do as humans, so we're kind of taking the shine off of consciousness?

   20. CPChamath Palihapitiya10:20

       I'm not sure it's that, but I would make two comments. I think this is a really important week, because it starts to show how fast the recursion is with AI. So in other technologies and in other breakthroughs, the recursive iterations took years, right? If you think about how long did we wait for, from iPhone 1 to iPhone 2? It was a year, right? We waited two years for the App Store. Everything was measured in years. Maybe things when they were really, really aggressive and really disruptive were measured in months. Except now these incredibly innovative breakthroughs are being measured in days and weeks. That's incredibly profound, and I think it has some really important implications to, like, the three big actors in this play, right? So it has, I think, huge implications to these companies. It's not clear to me how you start a company anymore. I don't understand why you would have a 40 or 50-person company to try to get to an MVP. I think you can do that with three or four people, and that has huge implications then to the second actor in this play, which are the investors and venture capitalists that typically fund this stuff, because all of our capital allocation models were always around writing $10 and $15 and $20 million checks and $100 million checks then $500 million checks into these businesses that absorbed tons of money. But the reality is, like, you know, you're looking at things like Midjourney and others that can scale to enormous size with very little capital, many of which can now be bootstrapped. So it takes really, really small amounts of money.... and so I think that's a huge implication. So for me, personally, I am looking at company formation being done in a totally different way.

   21. JCJason Calacanis12:20

       Yeah.

   22. CPChamath Palihapitiya12:20

       And our capital allocation model is totally wrong side. Look, fund four, for me, was one billion dollars. Does that make sense...

   23. JCJason Calacanis12:27

       Nope.

   24. CPChamath Palihapitiya12:27

       ... for the next three or four years? No, the right number may actually be $50 million invested over the next four years. I think the VC job is changing.

   25. JCJason Calacanis12:35

       Sure.

   26. CPChamath Palihapitiya12:35

       I think company startups are changing. I wanna remind you guys of one quick thing as a tangent. I had this meeting with Andrej Karpathy, I talked about this on the pod, where I said, I challenged him, I said, "Listen, the real goal should be to go and disrupt existing businesses using these tools, cutting out all the sales and marketing, right, and just delivering something." And I use the example of Stripe, disrupting Stripe by going to market with an equivalent product with one-tenth the number of employees, at one-tenth the cost.

   27. JCJason Calacanis13:04

       Yeah.

   28. CPChamath Palihapitiya13:04

       What's incredible is that this AutoGPT is the answer to that exact problem. Why? Because now if you are a young, industrious entrepreneur, if you look at any bloated organization that's building enterprise class software, you can string together a bunch of agents that will auto-construct everything you need to build a much, much cheaper product that then you can deploy for other agents to consume.

   29. JCJason Calacanis13:32

       Hm.

   30. CPChamath Palihapitiya13:32

       So you don't even need a sales team anymore. This is what I mean by this crazy recursion that's possible.
3. 23:57 – 37:38

   Generative AI's rapid impact on art, images, video, and eventually Hollywood

   1. JCJason Calacanis23:57

      Interestingly, when you talk about the emotion of making these decisions, if you look at Hollywood, I just interviewed on my other podcast the founder of-

   2. CPChamath Palihapitiya24:07

      You have another podcast?

   3. JCJason Calacanis24:08

      I do. It's called This Week in Startups. Thank you.

   4. CPChamath Palihapitiya24:10

      (laughs)

   5. JCJason Calacanis24:11

      70-

   6. CPChamath Palihapitiya24:11

      Successful?

   7. JCJason Calacanis24:11

      ... episodes. You've been on it four times, and it's millions of dollars a year.

   8. DSDavid Sacks24:12

      Please, please, uh, don't give them an excuse to plug it.

   9. JCJason Calacanis24:15

      Listen. I'm not gonna plug This Week in Startups, uh, available on Spotify-

   10. CPChamath Palihapitiya24:19

       (laughs)

   11. JCJason Calacanis24:19

       ... and iTunes and youtube.com/thisweekin.... Runway is the name of this company I interviewed. And what's fascinating about this is, he told me on Everything Everywhere All at Once, the award-winning film, they had seven visual effects people on it, and they were using his software. The late night shows like Colbert and stuff like that are using it. They are ruthless in terms of creating crazy visual effects now without ... And you can do text prompt to get video output. And it is quite reasonable what's coming out of it, but you can also train it on existing datasets. So they're going to be able to take something, Sax, like The Simpsons, or South Park, or Star Wars, or Marvel, take the entire corpus of the comic books and the movies and the TV shows, and then have people type in, "Have Iron Man do this, have Luke Skywalker do that," and it's going to output stuff. And I said, "Hey, when would this reach the, the level that the Mandalorian TV show is?" And he said within two years. Now, he's talking his own book, but it's quite possible that, that all these visual effects people, from Industrial Light & Magic on down, are going to be replaced with directors, Sax, who are currently using this technology to do ... What do they call the images, like that go with the script?

   12. DSDavid Sacks25:36

       Storyboards.

   13. JCJason Calacanis25:36

       Storyboards, thank you. They're doing storyboards in this right now.

   14. DSDavid Sacks25:40

       Right.

   15. JCJason Calacanis25:40

       The difference between the storyboard, Sax, and the output is closing in the next 30 months, I would say.

   16. DSDavid Sacks25:47

       Right.

   17. JCJason Calacanis25:47

       I mean, maybe you could speak to a little bit about the pace here, because that is the perfect ruthless example of ruthless AI. I mean, you could have the entire team at Industrial Light & Magics or Pixar be unnecessary this decade.

   18. DSDavid Sacks26:00

       Well, I mean, you see a bunch of the pieces already there. So you have stable diffusion, you have the ability to type in the image that you want, and it, it spits out, you know, a version of it or 10 different versions of it, and you can pick which one you want to go with. You have the ability to create characters. You have the ability to create voices. You have the ability to replicate a celebrity voice. The only thing that's not there yet, as far as I know, is the ability to take static images and string 'em together into a motion picture. But that seems like it's coming really soon. So yeah, in theory, you should be able to train the model where you just give it a screenplay, and it outputs essentially an animated movie. And then you should be able to fine-tune it by choosing the voices that you want and the characters that you want and, you know, and that kind of stuff. So yeah, I think we're close to it. Now, I think the, the question though is, you know, every nine, let's call it, of reliability is a big advancement. So yeah, it might be easy to get to 90% within two years, but it might take another two years to go from 90 to 99%. And then it might take another two years to get to 99.9 and so on. And so to actually get to the point where you're at this stage where you can release a theatrical quality movie, I'm sure it will take a lot longer than two years.

   19. JCJason Calacanis27:14

       It will, but look at this, Sax. I'm just going to show you one image. This is, the input was aerial drone footage of a mountain range, and this is what it came up with. Now, if you were watching TV in the '80s or '90s on a non-HDTV, this would look indistinguishable from anything you've seen. And so this is at a pace that's kind of crazy. There's also opportunity here, right, Friedberg? I mean, if we were to look at something like The Simpsons, which has gone on for 30 years, if young people watching The Simpsons could create their own scenarios or with AutoGPT, imagine you told The Simpsons stable diffusion instance, "Read what's happening in the news, have Bart Simpson respond to it, have the South Park characters parody whatever happened in the news today." You could have automated real-time episodes of South Park just being published onto some website.

   20. DSDavid Sacks28:07

       Before you move on, did you see the, the Wonder Studio demo? We can pull this one up, it's really cool.

   21. JCJason Calacanis28:13

       Yeah, please.

   22. DSDavid Sacks28:13

       This is a startup that's using this type of technology. And the way it works is you film a live action scene with a regular actor, but then you can just drag and drop an animated character onto it-

   23. JCJason Calacanis28:27

       Mm-hmm.

   24. DSDavid Sacks28:28

       ... and it then converts that scene into a movie with that character.

   25. JCJason Calacanis28:34

       Like Planet of the Apes or Lord of the Rings, right?

   26. DSDavid Sacks28:36

       Yeah, exactly. Yeah.

   27. JCJason Calacanis28:36

       Andy S- uh, Dakus? Was he the person who kept winning all the Oscars?

   28. DSDavid Sacks28:39

       So there it goes after the robot has replaced the human.

   29. JCJason Calacanis28:42

       Wow.

   30. DSDavid Sacks28:43

       You can imagine, like every piece of this just eventually gets swapped out with AI, right? Like, you should be able to tell the AI, "Give me a picture of a human leaving a building, like a Victorian era building in New York." And certainly it can give you a static image of that. So it's not that far to then give you a video of that, right? And so I, yeah, I think we're, we're pretty close for, let's call it, hobbyists or amateurs to be able to create pretty nice looking movies using these types of tools. But again, I think there's a, uh, a jump to get to the point where you're just altogether-
4. 37:38 – 1:12:35

   How to regulate AI?

   1. JCJason Calacanis37:38

      But let's pull up Chamath's tweet, "Of course, the dictator wants to dictate here." "All this incredible innovation is being made and a new hero has been born, Chamath Palihapitiya." A tweet that went viral, over 1.2 million views already. I'll read your tweet for the audience. "If you invent a novel drug, you need the government to vet and approve it (FDA) before you can commercialize it. If you invent a new mode of air travel, you need the government to vet and approve it (FAA) ." I'm just gonna edit this down a little bit. "If you create new security, you need the government to vet it and approve it (SEC) . More generally, when you create things with broad societal impact, positive and negative, the government creates a layer to review and approve it. AI will need such an oversight body. The FDA approval process seems the most credible and adaptable into a framework to understand how a model behaves and its counterfactual. Our political leaders need to get in front of this sooner rather than later and create some oversight before the eventual big avoidable mistakes happen and genies are let out of the bottle." Chamath, you really want the government to come in, and then when people build these tools, they have to submit them to the government to approve them? That's what you're saying here, and you want that to start now.

   2. CPChamath Palihapitiya38:49

      Here's the alternative. The alternative is gonna be the debacle that we know as Section 230. So, if you try to write a brittle piece of legislation or try to use old legislation to deal with something new, it's not gonna do a good job because technology advances way too quickly. And so if you look at the Section 230 example, where have we left ourselves? The politicians have a complete inability to pass a new framework to deal with social media, to deal with misinformation. And so now, we're all kind of guessing what a bunch of eight- 70 and 80-year-old Supreme Court justices will do in trying to rewrite technology law when they have to opine on Section 230. So, the point of that tweet was to lay the alternatives. There is no world in which this will be unregulated. And so I think the question to ask ourselves is, do we want a chance for a new body? So, the FDA is a perfect example why. Even though the FDA commissioner is appointed by the president, this is a quasi organization. It's still arm's length away. It has subject matter experts that they hire, and they have many pathways to approval. Some pathways take days, some pathways are months and years, some pathways are for breakthrough innovations, some pathways are for devices. So, they, they have a broad spectrum of ways of, of arbitrating what can be commercialized and what cannot. Otherwise, my prediction is we will have a very brittle law that will not work. It'll be like the Commerce Department and the FTC trying to gerrymander some old piece of legislation. And then what'll happen is it'll get escalated to the Supreme Court, and I think they are the last group of people who should be deciding on this incredibly important topic for society. So, what I have been advocating our leaders, and I will continue to do so, is don't try to ram this into an existing body. It is so important, it is worth creating a new organization like the FDA and having a framework that allows you to look at a model and look at the counterfactual, judge how good, how important, how disruptive it is, and then release it in the wild appropriately. Otherwise, I think you'll have these chaos GPT things scale infinitely, because again, as Friedberg said and as Sacks said, you're talking about one person that can create this chaos. Multiply that by every person that is an anarchist or every person that just wants to sow seeds of chaos, and I think it's gonna be all avoidable. I think regulating what software people can write is a near impossible task, number one. I think you can probably put rules and restrictions around commerce, right? That, that's certainly feasible, uh, in terms of how people can monetize. But in terms of writing and utilizing software, uh, it's gonna be as challenged as trying to monitor and demand oversight and regulation around how people write and use tools for, uh, for genome and biology, uh, exploration. Certainly, if you wanna take a product to market and sell a drug to people that can influence their body, you have to go get that approved. But in terms of, you know, doing your work in a lab, I- it's very difficult. I think the other challenge here is software can be written anywhere, it can be executed anywhere. And so if the US does try to regulate or does try to put the brakes on the development of tools where the US can have kind of a, a great economic benefit and a great economic interest-

   3. DFDavid Friedberg42:33

      ... there will be advances made elsewhere, without a doubt. And those markets and those, those places will benefit in a, in an extraordinarily out- ou- outpaced way. As we just mentioned, there's such extraordinary kind of economic gain to be realized here, that if we're not, if the United States, uh, is not, um, leading the world, we are gonna be following and we are gonna get disrupted. We are gonna lose an incredible amount of value and talent. And so any attempt at regulation or slowing down or telling people that they cannot do things when they can easily hop on a plane and go do it elsewhere, I think is, is fraught with peril.

   4. JCJason Calacanis43:10

      So you don't agree with regulation? Sacks, are you on board with the Chamath plan or are you on board with the free market plan?

   5. DFDavid Friedberg43:15

      Look, I'll, I'll say, I think, I think just like with computer hacking, it's illegal to break into someone else's computer.

   6. JCJason Calacanis43:20

      Sure.

   7. DFDavid Friedberg43:20

      It is illegal to steal someone's personal information. There are laws that are absolutely simple and obvious and, you know, no-nonsense laws. Those laws need to be implemented.

   8. JCJason Calacanis43:30

      It's not illegal to get rid of 100,000 jobs by making a piece of software, though.

   9. DFDavid Friedberg43:35

      That's right. And so I think try- trying to intentionalize how we do things versus intentionalizing, um, the things that we want to prohibit happening as an outcome, we can certainly try and prohibit the things that we want to happen as an outcome and pass laws and, and institute governing bodies with authority to oversee those laws with respect to things like stealing data.

   10. JCJason Calacanis43:56

       But you can jump on a plane and go do it in Mexico, Canada, or whatever region you get to. Sacks, where do you stand on this debate?

   11. DFDavid Friedberg44:01

       No, no, I'm, yeah, I'm saying, like, there are ways to protect people, there's ways to protect society about passing laws that, that make it illegal to do things as the output, as the outcome.

   12. CPChamath Palihapitiya44:09

       What law do you pass on ChaosGPT?

   13. JCJason Calacanis44:11

       Explain ChaosGPT. Uh, give an example, please.

   14. DFDavid Friedberg44:13

       Yeah. Do you wanna talk about it real quick?

   15. CPChamath Palihapitiya44:15

       It's a recursive agent that basically is trying to destroy itself.

   16. DFDavid Friedberg44:20

       Trying to destroy humanity.

   17. CPChamath Palihapitiya44:21

       Yeah, but I guess by first becoming all powerful and destroying humanity and then destroying itself.

   18. DFDavid Friedberg44:25

       Yeah. It's a tongue-in-cheek auto GPT. But it's not, it's not, it's not-

   19. CPChamath Palihapitiya44:31

       I mean, it's not, it's not, it's not a tongue-in-cheek auto GPT. It's a-

   20. DFDavid Friedberg44:32

       Well, the guy, the guy that created it, you know, put it out there and said, like, he's trying to show everyone, to your point, what intentionality could arise here, which is negative intentionality.

   21. CPChamath Palihapitiya44:41

       I think it's very naive for anybody to think that this is not equivalent to something that could cause harm to you. So for example, if the prompt is, hey, here is a security leak that we figured out in Windows, and so why don't you exploit it? So look, a hacker now has to be very technical. Today, with, with these auto GPTs, a hacker does not need to be technical at all. Exploit the zero-day exploit in Windows, hack into this plane, and bring it down. Oh, okay, the GPT will do it. So who is gonna tell you that those things are not allowed? Who's gonna actually vet that that wasn't allowed to be released in the wild? So for example, if you worked with Amazon and Google and Microsoft and said, "You're gonna have to run these things in a sandbox, and we're gonna have to observe the output before we allow it to run on actual bare metal in the wild," again, that seems like a reasonable thing, and it's super naive for people to think it's a free market so we should just be able to do what we want. This will end badly quickly. And when the first plane goes down and when the first fucking thing gets blown up, all of you guys will be like, "Oh, sorry."

   22. JCJason Calacanis45:47

       Sacks, pretty compelling example here by Chamath. Somebody puts out into the wild ChaosGPT, you can go do a Google search for it, and says, "Hey, what are the vulnerabilities to the electrical grid? Compile those and automate a series of attacks and write some code to probe those until we end success in this mission, you get 100 points and stars every time you do this."

   23. CPChamath Palihapitiya46:10

       Jason, it's such a, it's such a beautiful example, but it's even more nefarious. It is, hey, this is an enemy that's trying to hack our system, so you need to hack theirs and bring it down.

   24. JCJason Calacanis46:20

       Yes.

   25. CPChamath Palihapitiya46:20

       You know, like you can easily trick these GPTs.

   26. JCJason Calacanis46:23

       Right, they have no judgment.

   27. CPChamath Palihapitiya46:25

       They have no judgment. And as you said, they're ruthless in, in getting to the outcome.

   28. JCJason Calacanis46:30

       Right.

   29. CPChamath Palihapitiya46:31

       So why, why do we think all of a sudden this is not gonna happen?

   30. JCJason Calacanis46:34

       I mean, it's literally the science fiction example. You say, "Hey, listen, make sure no humans get cancer," and they're like, "Okay, well, the logical way to make sure no humans get cancer is to kill all the humans."
5. 1:12:35 – 1:16:37

   Bob Lee update, recent SF chaos

   1. JCJason Calacanis1:12:35

      All right, as we wrap here, what an amazing discussion. My Lord, I didn't, I never thought I would be...

   2. I wanna say one thing.

   3. Yes?

   4. We saw that someone was arrested for the murder of Bob Lee-

   5. That's what I was about to get to.

   6. ... uh, this morning. Yeah, which, uh, turns out that the report of the SFPD's arrest is that it's, uh, someone that he knew that also works in the tech industry, someone that none of us know-

   7. Possibly, right. So, so breaking news, yeah.

   8. Yes, possibly, but I, I, I wanna say two things. One, obviously based on this arrest and the storyline, it's quite different than what we all assumed it to be, which was some sort of homeless robbery type moment that-

   9. DFDavid Friedberg1:13:13

      ... has become all too commonplace in SF. It, it's a commentary for me on two things. One is how quick we all were to kind of judge and assume that, you know, a homeless robber-type person would do this in SF, which I think speaks to the condition in SF right now. Also speaks to our conditioning that, that we all kind of lacked or didn't even want to engage in a conversation that maybe this person was murdered by someone that they knew, because we wanted to kind of very quickly fill our own narrative about how bad SF is. And that's just something that I really felt when I read this this morning. I was like, man, like I didn't even consider the possibility that this guy was murdered by someone that he knew, because I am so enthralled right now by this narrative that SF is so bad, and it must be another data point that validates my point of view on SF. So, you know, I kind of want to just acknowledge that and acknowledge that we all kind of do that right now. But I do think it also does, in fact, unfortunately, speak to how bad things are in SF, because we all are, we've all had these experiences of feeling like we're in danger and under threat all the time when we're walking around in SF, uh, in so many parts of San Francisco, I should say, where things feel like they've gotten really bad. I think both things can be true, that we can kind of feel biased and fill our own narrative by kind of latching onto our assumption about what something tells us. But, but it also tells us quite a lot about what is going on in SF.

   10. JCJason Calacanis1:14:34

       Well-

   11. DFDavid Friedberg1:14:34

       So I, I just, I just wanted to make that point.

   12. JCJason Calacanis1:14:36

       ... in fairness, in fairness, and I think it's fine for you to make that point, I am extremely vigilant on this program to always say when something is breaking news, withhold judgment, whether it's the Trump case or Jussie Smollett or anything in between, January 6th, let's wait until we get all the facts. And in fact, quote from Sacks, "We don't know exactly what happened yet."

   13. DSDavid Sacks1:14:56

       Correct.

   14. JCJason Calacanis1:14:57

       Literally Sacks started with that.

   15. DSDavid Sacks1:14:59

       Yes.

   16. JCJason Calacanis1:14:59

       We do that every fucking time on this program. We know when there's breaking news to withhold judgment, but you can also know two things can be true. A tolerance for ambiguity is necessary.

   17. DFDavid Friedberg1:15:11

       Oh, I, I, I, I'm saying, but I'm saying I didn't even do that.

   18. JCJason Calacanis1:15:13

       But, but he didn't-

   19. DFDavid Friedberg1:15:13

       As soon as I heard this, I was like, I was like, "Oh, a homeless person's died there."

   20. JCJason Calacanis1:15:16

       That's a fine assumption to, but J-, but David, that is a fine assumption to make. That's a fine assumption to make.

   21. DSDavid Sacks1:15:20

       I think it's a logical assumption. Listen-

   22. JCJason Calacanis1:15:22

       You made that assumption for your own protection.

   23. DSDavid Sacks1:15:25

       (speaks at once) We got all these reporters who are basically propagandists trying to claim that crime is down in San Francisco. They're all basically seeking comment from me this morning, sending emails.

   24. JCJason Calacanis1:15:33

       Oh, god.

   25. DSDavid Sacks1:15:33

       They're trying to dunk on us because we basically talked about the Bob Lee case in that way. Listen, we said that we didn't know what happened, but if we were to bet, at least what I said, is I bet this case, it looks like, a lot like the Breonna Cupher case. That was logical. That's not conditioning or bias, that's logic. And you need to look at what else happened that week. Okay, so just the same week that Bob Lee was killed, let me give you three other examples of things that happened in Gotham City, AKA San Francisco. So number one, former fire commissioner Don Carmignani was beaten within an inch of his life by a group of homeless addicts in the marina. And one of them was interviewed in terms of why it happened, and basically, Don came down bit, from his mother's house and told them to move off his mother's front porch 'cause they were obstructing her ability to get in and out of her apartment. They interpreted that as disrespect, and they beat him with a tire iron or, or a metal pipe. And one of the hoodlums who was involved in this apparently admitted this. Yeah, play the video.

Episode duration: 1:33:06