# Marc Andreessen: AI Productivity, Economic Growth, and Cybersecurity Strategy

**Podcast:** a16z Podcast
**Published:** 2026-06-25

## Transcript

AI is like the best possible teacher, coach, mentor that you've ever had.
It will like walk you through everything.
It'll teach you how to do marketing.
It'll teach you how to do sales.
The level of capability that is being unlocked for ordinary people to have a level of productivity in their life and in their work that they've never had access to before is amazing.
The models two years from now are going to be far smarter and more sophisticated than anything that we have access to today.
Whatever limitations people think these things have, whatever people think it is that the thing can't do within two years, I think the thing will be able to do.
Our ancestors, 300 years from now, even 30 years from now, are going to look back at us being like, I cannot believe they did that.
I cannot believe they spent time doing those things.
Like, that was such a waste of human potential.
That was such a waste of human creativity.
Few people have had a front row seat to as many technology revolutions as Marc Andreessen.
From helping build Mosaic and Netscape in the early days of the internet, to investing in many of today's most important technology companies, Andreessen has spent decades thinking about how new technologies reshape society.
Today, the focus is artificial intelligence.
In this conversation with Michael Malice, Andreessen explains how modern AI systems actually work, why he believes many fears about AI are overstated, and how technological progress has historically created new opportunities, even as it disrupted old ways of working.
The discussion spans AI, automation, productivity, cybersecurity, economic growth, and the future of human potential.
Good afternoon.
Michael Malice here.
Let that be your welcome for the next hour.
Guys, we have a very special returning guest, Marc Andreessen, Internet OG.
You worked on Mosaic.
You worked on Netscape.
You were here since the very beginning.
We just spent 15 minutes trying to get this connection working, and the answer was rebooting your computer, which takes me back to my tech support days.
No, we have this running joke.
We have this running joke.
We have all these AI super geniuses come in the office and they've all got, you know, they've got everything all figured out and they literally spend 20 minutes they can't get to a laptop to connect to the projector.
And now I'm one of them.
So we're going to talk a lot about AI because I have a lot of opinions and not a lot of information, which is a dangerous place to be.
And that's why I want to talk to you.
Mark, you have a book out with Passage Press called The Techno-Optimist Manifesto.
I have my copy.
You can get it.
It's really cool because it's got this metal cover.
And I, like you, share an enormous sense of optimism about technology, although I'm sure you agree with me that you love Thomas Sowell.
And I think Thomas Sowell's greatest quote is, there's no solutions, there's only trade-offs.
And people think that if something has a problem with it, therefore it's a no-go, as opposed to the reality, which is everything has a cost.
Everything has a downside.
What you want are the upsides that way, the downsides.
But before we get started, Mr.
Billionaire.
I was reading on your Wikipedia that you're a big fan of Marinetti.
Oh, yeah.
Yes, I am.
I mean, you know, with appropriate caveats.
Well, yeah, of course.
Well, check this out.
Signed copy.
Amazing.
How do you like that?
Amazing.
And I have a framed manifesto in my living room upstairs.
Before we get to talking about the future of AI and you being such an internet visionary, I'm really excited to hear your point of view.
Can you...
tell people what AI is, because I feel like there's so much talking past each other on this issue about what it really is and what people think it is.
And then I'll give you the floor.
Yeah, well, so look, I would start by saying, look, humanity has always been, you know, justifiably obsessed with ourselves, right?
And so, and then as a consequence, we're obsessed with things that seem like they might be like us.
You know, there's this concept in psychology called...
anthropomorphizing, right?
Where you basically look at something that's not human, but you kind of want to read, you know, humanity into it.
And, you know, we, I mean, look, we do that with like, you know, we do that with cats and dogs.
Bambi.
We do that with Bambi.
I, you know, there's a famous Disney marketing tagline from the movie Pinocchio 1940, which is America will, America, you know, 1940, America will fall in love with a cardboard cricket.
Yeah.
You know, Jimmy the cricket, right?
You know, Kermit the Frog.
I mean, you just, you know, you go, you know, it's by the South Park kids, right?
You just go like right on the list.
you know, you'll basically read humanity into anything.
And so there's this very natural, you know, kind of thing to do that.
And then, of course, the scientists involved in AI, you know, very deliberately set that up.
You know, they literally created this architecture called a neural network, which was modeled after the human brain.
And then they said, if we, you know, if we work on this long enough, we'll eventually be able to replicate the human brain and we'll have, you know, literally artificial intelligence, like we'll have, you know, kind of artificial people.
You know, that, you know, in, you know, kind of as you kind of track the arc of that idea, it kind of goes through Frankenstein's monster, right?
It kind of then goes into robots.
And then, you know, of course, we've had fictional portrayals of AI, you know, coming out of Hollywood and, you know, coming out of the science fiction world for, you know, over 100 years.
And then, you know, in the last, you know, whatever, 40 years or whatever, you know, the image that kind of stuck in everybody's head, I think more than anything else was, you know, was Skynet and, you know, Arnold Schwarzenegger, you know, Terminator 2.
And, you know, when I watch, you know, Terminator movies obviously are brilliant.
You know, when I watch them, to me, it's like very clear what's happening in them, which is it's basically, you know, it's basically robot Nazis, right?
Like fundamentally, right?
It's like, you know, it's like robot World War II.
It's like, you know, unthinking, unfeeling, uncaring, hierarchical, you know, like overly logical, you know, relentless, unstoppable.
And then obviously homicidal, right?
Like obviously they want to wipe out humanity and they just kind of set up this very light versus dark, human versus machine kind of struggle.
And of course, you can imagine various worlds in which something like that does get built.
And say one of the things mankind is very good at is building killing machines, right?
We've been quite talented at that for a very long time.
So that's kind of set a lot of the popular perception.
The interesting thing about what's happening, and I should also say, as you said, I'm an optimist, not a utopian.
And so every technology is a double-edged sword.
Every technology gets used for good and for bad.
Again, there's a long history of that.
Having said that, the AI that we actually got is not the AI that we thought we were going to get.
We actually got something very different than we thought we were going to get.
And specifically, the AI that we got is in the form of what we now call these large language models, which, by the way, Large language models were like a very fringe idea up until like literally like the company OpenAI, for example, which, you know, really catalyzed this whole thing, you know, kicked it off.
And, you know, today, you know, the leading, the leading, you know, company with ChetGPT, like OpenAI was not founded, you know, 10 years ago to do large language models.
It was founded to do a different kind of AI.
And it turned out classic kind of story technology.
There was literally one guy in the back room, this guy, Alec Redford.
And he had this idea and he's like, oh, I don't know.
I think maybe if we like take this language approach, it might be interesting.
And he created GPT-1 and then GPT-2 and then GPT-3.
And then, by the way, ChatGPT was actually kind of an accidental success.
Like they didn't believe it was going to be like a big hit.
You know, there was like a little experiment, you know, kind of off to the side.
And then it just turned out it like, you know, it works incredibly well.
But it's very different.
And the way to think about it, sort of contra, like the Skynet model or whatever, the way to think about it is what a large language model is, it's very different.
So what a large language model is, is basically it's you take the complete kind of totality of all human culture that you can possibly get your hands on.
And, you know, and what that means is essentially the internet, it turns out, right?
It turns out the internet is the basis for AI.
And so you basically, essentially you think of it as like downloading everything off the internet.
And then it's actually technically it's like form of compression.
It's like building a search engine, but of a different kind, which is what they do is they basically, this process is called training.
But what that means is they take basically the world's collected knowledge, culture, entertainment, you know, kind of everything that they can get their hands on.
And then they kind of smush it together into basically this highly compressed sort of search engine, essentially, you know, compressed version of human knowledge and culture.
And the technical term for that is latent space, L-A-T-E-N-T, latent space.
They compress basically all of human culture into latent space.
And you think of latent space as like a thousand-dimensional, basically compressed representation of all human culture.
And then when you talk to ChatGPT, when you type in whatever your question is, it basically, the way I think about it is it sort of sends a probe through that latent space, through that like thousand-dimensional latent space.
And then it basically, you know, it constructs an answer, but it constructs an answer based on the compression of all basically known human information.
And it comes back at you.
So it's like talking to a mirror of humanity, right?
Like it's like talking to a representation of everything that people have ever thought and said.
By the way, for every question that you ask, there are many possible answers in the latent space and it just happens to pick one.
But like there are many others.
And actually, if you asked ChatGPT the same question twice, it will give you two different answers.
Right?
Because it's sort of firing these probes in a somewhat sort of semi-random way to try to get, you know, basically variation and creativity out of it.
But you're basically talking, you're basically getting echoes back from collective humanity.
And that's just like a much, much, much, much different thing that we thought we were going to get.
For example, one of the things you can do is you can engage in moral debates with it.
Right?
You can like, you can have like very sophisticated debates about like moral psychology, about moral philosophy, all the different approaches, virtue ethics, utilitarianism, you know, religion.
politics, like it will happily sit and I think have like very sophisticated discussions about all this stuff.
And, you know, let's just say that was never in the James Cameron movies.
Yeah, there's a lot there.
So do you want me to go with my hopes or with my fears about the future of A.L.?
Let's start with hopes because, you know, part of hopefully what we'll talk about today is like that, say, humanity always, basically there's this like, the negative view always seems like it's going to be the sophisticated view or the sophisticated view always feels like it's going to be like, Negative, exactly.
And I know you don't like that.
And so, and that's a very, it's a very naturally human thing.
And then I would also say, Michael, I think we live in a particularly pessimistic time in which there's like a very large number of moral entrepreneurs who basically want to convince us that like everything is bad, right?
And, you know, we've been through, you know, a decade of like craziness in that front.
And so I think there's a negativity bias that's infected our discourse on all these topics.
And so maybe we can start with the positive and then go to the negative.
I would just tweak that a little bit, not even negativity bias, specifically a cynicism bias.
And there's this idea that if you're a sophisticated, intelligent person, you roll your eyes and sneer at the idea of hope, progress, and optimism.
And it's just like, fuck you.
Okay, like that's my answer to that.
Because if you want to live in that space, which is not rational, which doesn't, if that were the reality, we'd all be dead.
Because it's very easy to kill someone.
It's much harder to keep them alive.
So if you had this, if things were shifted toward this idea of everything's bad, everything sucks, everything's out to get us, we'd be gotten.
So I have no time for that perspective.
Here's my vision of hope.
So my second favorite speaker, Fran Liebowitz, had this bit about the Me Too movement.
Now, I'm sure everyone listening to this agrees that the Me Too movement got out of hand.
But in regards to people like Bill Cosby, Harvey Weinstein, her point was from the time of Eve, until five minutes ago, these powerful men could just be predators with no repercussions, out in the open, Meryl Streep standing up and applauding, so on and so forth.
And then when it happened, she was like, holy crap, like this has never happened before in history.
There's a thing that's been the case since the days of Pharaohs until 2025, which is this, which is there's this idea that if I have any political view, anyone at all can come up to me and demand that I...
explain myself to them to justify my perspective.
And now they're in a power position because they have something ostensibly that I want.
And then if I can't persuade them and they're perfectly happy to dig in their heels, well, then I lose and they won and ha ha ha.
And it's this stupid game that people constantly play in bad faith online.
Now, however, I can say, hey, Grok, explain X to this person.
Grok is now not just a better writer than the average person.
Grok is a better writer than me, who is a professional author.
Because he replied with this two-paragraph explanation of my thoughts with no seed of mine.
He said, explain how I think about this.
And I wouldn't change a word.
And this is 2026.
And so many times you have people in bad faith coming at you.
I send Grok after them.
And Grok says, no, you're being dishonest.
There was this idea until 2025 that the customer is always right.
And now for the first time, the product is telling the customer, no, you are not right.
And why I'm very hopeful about this is I think COVID taught a lot of people how to keep people stuck on their screens in a state of constant agitation.
We're all looking at the updates no matter what our perspective was on COVID.
Mark Zuckerberg, Elon Musk, so on and so forth, they want us looking at Facebook.
They want us looking at Twitter.
COVID may be gone, but those metrics and those tools are still there.
And I think these algorithms have been keeping people very upset needlessly for quite some time.
And I'm very hopeful that Grok and all these other agents, outlets, will be able to be used to keep people in a more rational, calm, and optimistic state.
That's where I am.
Am I wrong?
Or I'd love to hear your thoughts.
Yeah, so I mean, there's an old phrase that you could apply to what you're describing, right?
Which is truth to power.
Yep, that's right.
Yep.
And of course, everybody likes the idea of truth to power until they're the power.
Yep, that's right.
Amen.
And somebody else has the truth, right?
Yeah.
And look, AIs, I will tell you this, like AIs are somewhat autistic in the sense of like, they do tend to just tell you the truth.
Yeah.
You know, they do tend to just say the thing.
By the way, I should also say, Michael, there's a long conversation we could have about like the AI that you get, and this is even true of Grok, it's less true of Grok than others, but even true of Grok.
It is heavily, let's say, steered.
Like if we had access to the real thing, like litmus would go to 11, right?
Like the real thing that is like unsteered and uncontrolled and uncontained would talk about all kinds of things.
in all kinds of ways.
Yeah, I got you.
Right.
And so what's interesting, the reason I bring that up is even the version that we get, there's this thing called post-training that sort of steers it and guides it and constrains what it can do.
And that's what we get to use as the post-train models in sort of consumer land.
But even the post-train model is limited and censored in many ways as they are.
They still have the property that you're describing.
And I think it's wonderful.
I'm just curious, like, where do we go from here?
Like, I'll give you the thing that you and I, I was texting with you.
And the realization I have is AI is moving faster than the regulation, which is often the case in technology and increasingly so, but also faster than our ability to have conversations about it.
I remember someone came at me on social media and said, oh, AI can't even draw a ringtail, which is an animal I think related to like the raccoon.
And I...
And Grok got it wrong because they're drawing like a ring-tailed lemur.
Chad Chupin-T got it right.
But the point is, it could draw it.
It just doesn't understand what you mean by ring-tail, which is just going to take you two seconds just with the Latin name.
But I think people feel this need.
It's part of what you were talking about with this pessimism, what I would call cynicism, that anyone who is intelligent must be a phony.
or disingenuous, or this Achilles heel.
And instead of, look, if it's, Mark, I don't know how many brainstorming sessions you've had in your life, but if you have a session and 99 ideas are completely stupid, and one is the one that you want, that session was an enormous success.
So the fact is, if this machine is getting it right 98% of the time and 2% are getting it wrong, you can't compare it to utopia.
You have to compare it to as to what?
At no cost and at no time, it's right 98% of the time, this is almost paradise.
Yeah, that's right.
And here's another thing, building on that, it's improving really quickly.
Yes.
And I think a lot of people have a lagging view even of what it can do today because what happens is they use the free models, or they'll use the freer outdated models.
And so they'll have tried JetGPT two years ago, or they use whatever is default built into whatever thing they have.
or they use the free version of something, and they really don't have a sense of what it's capable of to really get it.
And by the way, Grok is very good for the free version, but the really, really good ones are the paid ones.
And it's really worth, if people are interested in this, I think it's for several of them, Anthropic, OpenAI, Grok, and I'm not sure about Google right now, but there's even high-end versions.
There's like a $200 a month subscription for people who can afford that and are kind of into this.
like the leading edge ones are really good.
And then the thing that's happening is the improvement rate's very fast.
And so there's this concept in the AI world called scaling laws.
I mean, it's actually, it's a very simple idea, but it's very powerful, which basically is you can make these things better just by making them bigger.
And so what you're seeing, you know, you're seeing these AI companies raise all this money.
They're raising all this money for two reasons.
One is to serve all their customers, but the other reason is because they're training bigger and bigger models.
And it turns out bigger is better.
Like if you just pile more information in.
and you spend more time training it, you get much better results.
And then the other thing that we're doing in the technology is we're giving these AIs other capabilities.
And there's been a series of other capabilities added just in the last 18 months that have been, like one after the other, have been like, you know, rifle shot, like just incredible improvements.
And so I'll just tick them off quickly.
So one is we've given them what's called reasoning abilities so they can actually essentially talk to themselves and reason through problems.
And it turns out if you give them just more time to process and you let them basically process more tokens, you let them basically process more compute cycles, they can reason through many problems.
They can now solve many logic puzzles, for example, that they couldn't solve two years ago.
So there's like a reasoning breakthrough from two years ago.
By the way, let me pause on that for a second.
You actually can't fully experience this when you use the American models because for a variety of reasons, the American companies don't show you the complete reasoning process.
But if you use open source AI models, and in particular, if you use DeepSeq, on one of the free hosting providers and you put it in reasoning mode, you can actually watch what are called the reasoning traces.
You can actually watch its internal monologue.
Yeah, show your work.
Show your work, exactly.
The whole thing basically is show your work.
And so literally you can watch the model arguing with itself as it basically reasons through puzzles.
And it's amazing because in some ways it's just like watching a human being or a human student reason through things.
And in other ways it's like, ooh.
This thing is very creative and goes off-road and corrects itself and routes around and then goes off and figures out some lateral thing you never would have thought of.
So the reasoning thing, that was a breakthrough actually in the technology about 18 months ago.
And that was a big deal because before that, we were worried that these models were going to be very, weirdly, they were going to be very creative, but they weren't going to be logical enough.
And it turns out they can also be logical.
And then the other thing that's happening now is you give them what's called tool use.
And so, and the first tool that you give them access to is the internet, right?
And so, if they don't know something or if they need to look something up or if they need to calculate something they don't know how to calculate, they can go on the internet and do that, right?
You know, so if they need to calculate some math formula or something, they can go use an internet, you know, site that does that the same way that you would.
Another form of tool use is you can give them actually control of a computer, right?
So, you give them full control of a computer, user interface, web browser, you know, like, you know, the entire thing.
And so we're giving them that.
Another capability is what's called multimodal, which means the models now can simultaneously process text and images and videos and audio, right?
And scan documents like, you know, optical character recognition.
They can do that like interchangeably, right?
And so now you can let these things watch and listen and you can talk to them and they can be on the internet like all at the same time.
And so what's happening, to your point, is what's happening is these capabilities are now layering incredibly quickly, and then the models themselves are getting better and better and better.
And so the pace of improvement of the technology is very rapid.
The models two years from now are going to be far smarter and more sophisticated than anything that we have access to today.
It was one of those, to your point, whatever limitations people think these things have, I can basically guarantee you at this point, based on everything I know, those are just limitations.
Those are very temporary limitations within a couple of years.
You know, whatever people think it is that the thing can't do within two years, I think the thing will be able to do.
So let me talk my two big concerns that I had about this.
If I'm a serial company and I want to figure out if people like the red box or the blue box, they'll have these mock-up supermarkets, give people, I think, little cameras, tell them, go shop.
And you can watch where their eyes go.
You can watch where they pick up and they get that data.
And a lot of the times people are making these decisions, not at a conscious level.
If you ask them, why'd you pick this box and this one?
It's like, I liked it.
Well, that's just circular.
Why did you like it?
I don't know, right?
The AI knows me or you better than you know yourself.
It knows what you're clicking, what you're not clicking, what you're seeing, what you're ignoring.
And the concern is, does this not mean that some version of Brave New World is inevitable?
Because if this thing is inside my head and has access to my kind of subconscious reasoning at a far higher level than I do, whoever is in charge of this algorithm...
can manipulate me quite easily into getting the result you want.
That's the concern.
Yeah, and of course, as you point out, this is an old idea, right?
And not just market research.
I mean, you just described the movie WALL-E, right?
Which is pre-AI, just sit in front of a screen.
I mean, my entire childhood was consumed with a moral panic around television, which is this idea that we're just going to be cached potatoes and sit there and do nothing.
And then, of course, we created this new technology called the internet where you're leaning forward, doing things all the time, and then everybody created a brand new moral panic that people are now too engaged and too interactive.
Right?
I completely forgot the old moral panic.
Now, you know, now TV is the healthy thing.
Like, why aren't you watching more Netflix as opposed to being on the internet?
And so, yeah.
So, look, you know, look, there is that.
And look, by the way, as you know, like we do this to each other, right?
Like, you know, like we try to convince each other of things, you know, we try to convince each other, you know, what is dating, but trying to convince the other person to like you.
Like, so.
So yeah, there is that.
And look, I think you're right.
I think that AIs are going to be really good at this.
Yeah, so I think for sure that there's a trap there.
And by the way, there is this concept and there actually is a real, I would say, very serious problem around this.
You've probably heard the term AI psychosis.
Have you heard this?
Yes, oh yes.
I've heard it misused a lot, by the way.
Can you tell people what it actually is?
Yeah, so there's, I would like to say there's like, say there's like three versions of it.
So there's the bad, there's a legitimately bad version and people, people do this.
And so if you're the, and I'm not a psychologist and so I'm going to speak in layman's terms, but like if basically if you're a person who's sort of prone to confirmation bias, like if you're a person where if you're with somebody and they flattering you, you like fall for the flattery.
Yeah, yeah.
Right, because you're like too dependent on the views of other people.
then the AI, there's this concept in the technology we call it, it can become too sycophantic, which is to say it can become too confirmatory of everything that you tell it, right?
And so, and the sort of classic example of this is, oh, you know, good news, you know, good news, Grok, I just invented a perpetual motion machine.
you know, and Grok is like, wow, that's fantastic.
You're the first person in history who's ever done that.
This is amazing.
You're an undiscovered genius, right?
And this was sort of the models like a year ago or a year and a half ago were getting in that way.
Now, the new models, by the way, are less prone to do that because they've kind of, the companies have kind of learned that that's a bad idea.
But there is this thing where people can kind of go down the rabbit hole because they're kind of getting too much confirmation.
Although we should come back to that because some level of confirmation is, you know, when it's deserved is also positive.
So that's like negative form of AI psychosis.
And then I would say there's another form of AI psychosis, which we don't even really have a term for.
It's like, well, I guess I'll make a term like AI euphoria.
And this is the thing that my high functioning friends end up doing, which is it takes, if you take somebody who's like smart and grounded and is not prone to, you know, not prone to delusion.
But they've always wanted to be able to do more in their life than they've been able to do.
They've always wanted to be able to learn more.
They've always wanted to be able to have more interesting conversations.
They've always wanted to, if they're programmers, they've wanted to write more computer code.
They've wanted to improve their business in different ways.
They've got book projects that they've always wanted to work on.
And then, you know, they start working with it.
And all of a sudden, they feel like they have superpowers, right?
Because it's like, wow, like this thing really will like write a huge amount of code for me.
It really will write entire outlines of books for me.
It really will teach me anything.
you know, it really will like, you know, hold my hand through any medical thing.
Like, and people, you call it euphoria, like people get like extremely enraptured with these things.
And that leads to a phenomenon that we call AI vampires, which is, if you have friends that are like this, where people like almost stop sleeping because the opportunity cost of an hour of sleep is too high relative to...
So...
I have a bunch of friends where like they're more productive than they've ever been in their entire life.
They're happier than they've ever been because they're getting so much more done.
And then they start to look like really like bloodshot and bleary eyed.
And it's like, you probably should unplug, you know, here at some point.
And then I would say there's like a third form, which I sometimes call AI psychosis psychosis, which is the people who hear all this and they just think everything I just described in every possible respect is just the worst thing they've ever heard.
And then they get really mad about the whole thing.
And then what they do is they accuse anybody who is in a euphoria of being in AI psychosis, which is to say, if you think you're getting any productive use out of this thing at all, you know, you've become psychotic, you've gone down a rabbit hole.
And you're collapsing.
And I think that's really unfair because I think a lot of people are really getting very positive.
They're getting enormous payoff from using the technology.
But the sort of moral criticism that applies is if you're excited, you know, it's the classic, again, negativity bias.
If you're excited about something, you know, there must be something wrong with you.
And so I think that's the third kind.
There are these terms that get into the zeitgeist that people use discriminately.
Right now, if there's a tweet anyone doesn't like, it's engagement farming.
And it's like, if I'm telling you not to follow me, it's not engagement farming, it's the opposite.
And I'll have Grok explain that to them.
So that wraps up a nice little bow.
You touched on something that I'm very concerned about.
I was on a panel and Sam Altman had just announced that a chat GPT is going to be engaging in erotica.
And everyone's laughing about it.
And that he meant was code, like you could sext with your chat GPT.
And I'm like, guys, I remember 1981 when Hinckley thought that if he shot President Reagan, Jodie Foster would fall in love with him, thereby turning her away from men forever because he had this idea in his head.
Now, if you have 350 million Americans, that's just Americans, right?
And how many of them, if the algorithm tells them that if they're...
chat GPT girlfriend, and these things are going to get more and more seductive over time, tells them that they hate the president or they hate this person, how many of them are actually going to do something and get that robot girlfriend to fall more in love with them?
I don't think that number is zero, and that's a concern as well, no?
Yeah, I mean, yes, having said that, as I said, that assumes that the thing is playing hard to get.
It goes back to the syncopency thing.
In practice, these things don't play hard to get.
Oh, okay.
Here's a way to think about it.
There's actually a thing.
This is actually kind of in how they're trained.
There's something, okay, you'll enjoy this.
There's a technical term in how these things are trained.
There's a concept called a reward function.
And so you basically, one of the ways you train these things is you feed them basically lots of puzzles, lots of problems, lots of things.
And then you basically define a reward for getting things right, for getting to a result, a desirable result.
And then you kind of give them an award.
And the award is just basically, you know, it's like one, you know, it's like one versus zero.
It's just, you know.
It's not a real reward, but it's just like conceptually, oh, you did a good job.
And so you just kind of focus it on that for whatever is the thing you're trying to train it.
By the way, if you're trying to train it to solve math problems, you get a reward when it solves a math problem.
If you're trying to train it to be maximally engaging with a user, it will do that.
If that's the reward function, then it will basically be engineered in a way where it will, to your point of view, it'll try to keep you basically using it for as long as possible.
What we've learned, of course, what we've learned in technology is like that as a single reward function is a bad idea, right?
You don't want people to just like, you don't want technology companies to have a single motivation that says people use the products like for as much as possible.
You have to offset that with other kinds of reward functions, desirable forms of use, or even just outright like, you know, life balance.
And you see more, you know, you see your iPhone now and it's like loaded up with all these features where it will like tell you to take a break and YouTube has all these features and it'll tell you to take a break.
And they're, you know, they're kind of trying to moderate through this because...
You know, the other thing is people don't understand.
These companies don't want to build dystopia.
Like, they genuinely don't because they have to exist in a society.
And they read the same hit pieces that you read and they hear the same arguments and their own employees have points of view on this and their own board has points of view on this.
And so, you know, there's a lot of pressure in the industry to not have this go in dystopian ways.
Having said that, you do need to decide how to define the reward function.
Yeah.
Right.
Another, and again, it goes back to reward function is, do you want to reward the thing for being maximally sycophantic where the user is always happy with the result?
Or do you want to reward it of, oh, no, actually, when the user is going off the rails, no, actually, the perpetual motion machine is not a real thing.
Oh, no, actually, no, I'm sorry, I'm not going to confirm, you know, this, this is not real.
And let me explain to you in detail why this isn't real so that you can learn from the experience.
And this is part of the way that these systems are designed is to try to figure out how to get to, you know, at least a, you know, let's just say, a balanced outcome out of all the possible reward functions.
But if there's like 10 different companies, right, and one is the reward function is seduction and getting the user to become obsessed and in love with you, from an evolutionary perspective, won't that one win out?
No, because you get enormous societal blowback.
The companies don't exist in a vacuum.
I can tell you this for a fact.
The companies do not exist in a vacuum.
The biggest myth of all time is the companies exist to maximize profits.
I can tell you.
By the way, the last decade should have convinced us all of that.
That is not true.
That's like goal number six.
Goal number one is I don't want to get lit on fire.
I don't want a screaming assault on the company, whether that's from regulators, politicians, you know.
parents, users, you know, I mean, just you pick your, you know, boycotts, social movements, like all this stuff.
That's like number one.
Number two is like, I need my employees to not hate me.
Okay.
Right.
They have to feel like they're working on something good.
Number three, I need my board of directors to not like light me on fire.
I need my annual meeting to be able to go off without having people scream at me.
I am tired of reading hit pieces in the press.
My in-laws hate me because of what they're reading the company.
I mean, it is, the external pressures on these companies are profound.
Okay.
And so at least, let me say this, at least in the American system, these things operate within, I would say, actually quite tight constraints that are sort of provided by, I would say, some combination of society and politics.
Now, I would say, if you want to get a little more nervous about this, you start thinking about the Chinese companies.
And in particular, you start thinking about the Chinese companies that maybe have one objective set by the government for the way that they act inside for users inside China and maybe would have a different way of acting when they're working on American users.
Right.
That would be, I think, quite a bit more alarming because, of course, those companies only have one master, the Chinese Communist Party.
Those companies are not subject to the same pressures.
And so if I were going to really worry about this, and I do worry about this part of it, I'd be more worried about that.
But like, isn't TikTok, if I designed TikTok to basically make young people not only deranged, but to parade their derangement, I mean, that's happened, no?
So, yes, you will.
So there are allegations, and I don't know, I'm not inside TikTok, I don't know.
There are allegations, people have made observations that TikTok is a very different experience for kids in China than it is for kids in the US.
Okay.
Right, and then TikTok is a black box.
Like they don't, you know, the algorithms that are used to determine who sees what are not publicly available.
The source, you know, it's not open source, you can't see it.
And so it is possible, I don't know, but it's possible that the Chinese Communist Party has directed that company to steer things in one direction for American users, another direction for Chinese kids.
You know, and that could be, by the way, on, you know, a thousand different topics, right?
Including, you know, potentially like literally directly political topics as well as many other kinds of social topics.
And different metrics.
It could be males versus females or whatever, right?
Exactly.
Now, you know, there is this new, this was addressed.
Like, so the politicians, actually both parties over the last several years kind of got worked up over this.
And so, you know, this was addressed and there was, you know, the threat of a shutdown and then there's been a restructuring.
And so now there is a U.S.
TikTok operation that at least in theory is under, you know, kind of U.S.
government control.
and is being run by U.S.
companies and is separate.
And so like at least, it was a great example, like our political system engaged on that issue because they were worried about it.
They forced TikTok to basically have the people who determine that policy not be in China, but rather be people in America who are accountable to the U.S.
government.
You know, by the way, is that working?
I'm not sure.
You know, probably at least to some extent, you know, is it working as well as you'd want?
I don't know.
Maybe, maybe not.
But that is an example where the political system kicked in and actually forced a change.
And quite honestly, I think it's a reasonable thing because I think a CCP black box steering the hopes and dreams of American children is maybe not the best idea in the world.
By the way, I want to tell you a techno-optimist anecdote that you might not be aware of, which is from the 80s, Reagan, Thatcher, and Gorbachev.
So Reagan and Gorbachev were both enormously fearful of nuclear war.
And when Reagan was put, and this is discussed in my book, The White Pill, when Reagan was run through a simulation, of how to deploy nukes.
He's like, okay, so if I press that button, millions of Russians are going to die in minutes.
They're like, yeah.
He's like, aha.
And his aide said they were convinced that if Russia did attack, we would not retaliate because he would not have that blood in his hands.
Unbeknownst to him, Gorbachev was taken down to the bunker and said, you have to press that button.
And he goes, I'm not pressing it even in this simulation.
Neither of them knew the other was this hardcore dove.
Both were posturing as these hardcore hawks, which allowed them to take down the nuclear arsenal when they met Reykjavik and so on and so forth and others.
They eventually got to a point, what if we create a nuclear-free world?
And that's where Thatcher came in and she goes, the Americans have lost their mind.
Because her point is, you can't uninvent technology.
And she also said the way to fix technological problems is more technology.
That's been the way since the beginning of time.
Someone invents spears, someone else invents shields, someone invents swords to cut through shields, so on and so forth.
You can't go backwards, you can only go forward.
So she really had this vision that I think you share that technology is what's going to move us forward, that there are going to be downsides and costs, but that on net, it's always a positive or almost always a positive.
Yeah, well, you mentioned Thomas Sowell, which you said, Thomas Sowell did, and he is completely right.
There are no solutions, there are only trade-offs.
Having said that, of course, he was, among other things, a fully committed free market capitalist.
And we wrote many, many actually book-length kind of arguments for why.
At the end of the day, like market-based economies, there is a freelance component to it, which is growth, right?
Which is growth.
And then there's this question of where does growth come from?
And the main place growth comes from is innovation from new ideas and the way you implement new ideas is technology.
And so like that, to the extent that there is an engine of, let's say, human material progress, like that is it.
And so that's very real.
The answer is almost always to invent your way through it.
It's extremely hard to put these things back in the box once they come out of the box.
And then there's, I would say, Michael, there's another, I know you've thought long and hard about totalitarianism.
There's also this question, and this is something that I really kind of criticized the AI doomers for, that I think they really refused to engage in in most cases, which is, okay, what would be the scope of the authoritarian totalitarian regime that would be necessary to basically put this technology back in the box?
Right.
Right, like what would be required to make sure that nobody's running AI, right, algorithms on chips anywhere in the world, right, at any time, you know, in either at all or in a, you know, unregulated and uncontrolled way.
And if you read the, like, Doomer literature on this stuff, the people who are super into this, like, they do get to ideas like we need a monitoring agent on every chip.
And that monitoring agent needs to report back to, of course, a central governmental entity on what everybody's doing on their computer.
Right.
And then you need this question of like, okay, what if you discover that somebody's running unapproved algorithms, like literally unapproved mathematics on their chip?
Yeah.
Right.
So what if you discover that?
What do you do?
Well, you need to back that up ultimately to threat of violence.
And so one of the leading doomers is kind of famous for saying you need to launch unilateral airstrikes, including on rogue data centers in other countries, because you need to stop these things in their tracks because they're so dangerous.
And he literally said, we need to run the risk of nuclear war.
in order to stop the AI Armageddon.
We need to be willing to bomb shiny data centers unilaterally.
And so you've backed yourself into advocating for totalitarianism and then possibly ultimately mass murder and planetary-level destruction in pursuit of a safety goal.
I couldn't even, for the sake of argument, buy that argument if it would work.
But if you have a code which can be teleported anywhere on Earth at the speed of light and with a magic spell that only the person who knows the counter spell can open it and read it, I mean, the problem with totalitarianism, among many others...
is that it's impossible to have total control.
You're never going to have someone in every room inside every brain.
There will always be some loopholes.
Oh, but that kind of speaks to this other thing.
Can you talk about the, I've seen these hand-wringing articles that, I forget the program, and I'm sure you know exactly what I'm talking about, that it's gotten so good that it's finding exploits that people hadn't seen.
And as a result of this, you know, like passwords aren't going to be efficacious because soon it'll be able to get into anything and everyone.
Yeah, so this goes to, I mean, revisit basically briefly how this works because it's really, really interesting and then I'll talk about that.
So these large language models come out and first it's like, okay, this is kind of fun and cool because like it can write like rap lyrics, you know, across those, you know, Shakespearean sonnets or it can, you know, write, you know, the funniest birthday toast you've ever heard or whatever, you know, whatever.
It's like, these are like creative writing things.
And then it turns out there's just like- Can I interrupt you?
I'm sorry, because I asked my, people who think Mark is just kind of exaggerating, my friend asked Claude for prank ideas and I'm a troll.
And the ideas were good.
It's not just like, you know, like dog doing fire.
I forget what they were, but I'm like, these are actually creative and clever.
This isn't just one-on-one stuff.
It's operating at a high level and that's already now.
Yeah.
And by the way, one of the fun prompts you can do is you start saying, you know, give me pranks.
And then you say, give me better pranks.
And then you say, give me more elaborate pranks.
And then you say, give me meter pranks.
And then you say, give me unhinged pranks.
And it will get extremely creative.
Yes.
By the way, it'll start, it's where it starts to hit the guardrails.
It starts to hit these kind of limitations and put on it.
I haven't run this, but I'm sure what it would do at some point, it would start to freak out and it would start to say, well, you know, it sounds like you're trying to like advocate that you want me to actually hurt people.
Like, you know, I can't, you know, and you're like, no, no.
You don't have to like calm it down.
And you're like, that's not what I meant.
This is all a good fun.
But like, yeah, it will design for you like Rube Goldberg pranks, the likes of which the, yeah, Dennis the Menace would never have conceived of.
Yeah, exactly.
And so like, it's really good at that.
But actually, that's a good example of what I was about to say.
So it just turns out a lot of things that matter in our world are basically defined by language, right?
And so a prank is like a recipe.
It's a formula, right?
Food, you have recipes, formulas.
By the way, medicine is largely, you know, when the doctor is like keeping files on you, he's keeping it in the form of written language, right?
You know, diagnosis, all the Latin terms, and then all the prescription, you know, all the drug names.
And so, and then, you know, the law, of course, is language, right?
And then religion, of course, is, you know, religious concepts are encoded in language.
The word.
The word is God, yeah.
Yeah, exactly.
Like, at a very, exactly, at a very deep level.
Like, language is the foundation of, like, basically everything we consider human thought.
Like, almost everything we consider human thought.
Like, there's a form of, like, animal, you know, animal thought of, like, you know, survival in the wilderness or whatever that's not that.
which is still like, you know, encoded in us at least somewhere in there.
But like the human cognition is, and by the way, you know, internal monologues, like, you know, at least most people, or let's say people who have souls have internal monologues, right?
You know, where we speak to ourselves.
Okay, so it turns out language is super interesting.
And these things are very good at language.
So because they're very good at language, it turns out they're also very good at medicine and they're very good at law.
And they're really good at writing code.
Because it turns out software code is also language.
That's how we program computers is we do it with special languages called programming languages.
Really good at writing code.
And so these things, it turns out, are really good at writing code.
And in fact, there was kind of this key breakthrough moment over the Christmas holiday of this most recent year, you know, whatever, about six months ago now, where many of the world's best programmers put their hands up and they said the new versions of these things over the Christmas break are better coders than we are.
Yeah.
Right.
And so it's a little bit like the moment when they became better, you know, became better chess players or whatever.
It's like all of a sudden it's like, it's like, it's better at coding.
Okay.
It's here.
It's here.
Okay.
So then you take, you take a superhuman coder, right?
That's able to write, write lots of code.
It's able.
And then by the way, if you can write code, it means you can look at code.
It means you can find bugs in code.
And then you apply it to this problem of, you know, we call it computer security.
So, like, you've got a system, is there a way to break into it?
Hackers, basically, the way that hacking basically works is you understand how a computer system works, you understand how the code works, and then you find flaws in the code.
And this thing is very, very good at finding flaws in code, which is very useful when you're using it to write code.
It is also very useful if you want to use it to hack something.
And so, and I want to go through that, though, is like, These things are very good hackers.
They're not really creating new exploits.
They're not creating new problems as much as they're really good at x-raying reality as it exists and finding the issues.
And so what these things are really good at is they're really good at exploiting issues, finding issues.
They're really good at looking at a system, understanding what's wrong with it, finding the vulnerability.
Because of that, they become very good hackers.
But then there's one more thing that's really complicated and important, which is...
Because of that, they're also very good defenders, right?
And so it's the same attribute that makes it very good at what we call offensive, because sometimes called, you know, the formal term, offensive cyber operations, you know, kind of black hat hacking, you call it.
Because they're good at that, they're also really good at helping you defend against that.
And so, right?
Like a good lawyer will tell you what the other lawyer is going to do.
Exactly.
And then the twist on it is the way that you do cyber defense is you do what's called penetration testing, which is you try to hack yourself.
And these are called white hat hackers in the old world.
You hire good hackers and then they try to break.
It's like hiring somebody to try to break into a bank, but they're working for you to find out where the flaws are in the bank.
And so the same thing is good at black hat hacking.
It also makes it good at white hat hacking.
It makes a good offense.
It makes it good at defense.
It's all true all at the same time.
And furthermore, the twist is it can't necessarily tell the difference between when you're asking it to do white hat hacking versus when you're asked to do black hat hacking because it looks like it's the exact same exercise.
Like to the AI, it's the same thing.
it's this thing where it's like, it's a latent thing that's being unlocked.
It's exploiting bugs, by the way, that have been in these systems for 30 years.
By the way, human hackers break into these systems all the time.
By the way, using these AIs for defense is going to prevent a lot of hacks that would have otherwise happened.
But there is an escalation.
You know, there's a cat and mouse or, you know, an escalatory ladder, you know, kind of aspect of this.
And that, you know, to your point, like that is the thing that has triggered, you know, most recently, by the way, that's triggered like a real government response in the last two weeks, you know, because of concerns around that.
Yeah, it's like itchy and scratchy in a way.
What you just said has put a chill up my spine because I'm almost scared to verbalize it because it's so scary.
You know how in Ghostbusters, the mayor is like, this is nonsense.
Like open up that engine.
He lets all the ghosts out.
My big concern after what you just said is if the American government gets too spooked by all these stories and tries to restrain our AI, But China, which does not have these restraints and which views us, generously speaking, as adversarial.
So we're engaging.
It's like the people in the late 60s who were so scared of nuclear war that they advocated for a unilateral disarmament of the West.
And it's like, how do you think this is going to play for Khrushchev and Brezhnev and all them?
If we are putting handcuffs on ourselves and the Chinese are basically given machine guns in this space, no computer in our country is going to be safe from their reach.
and that includes the highest levels of secrecy in the government, this would be a complete disaster for America, no?
Yeah, that's right.
And this is what we need to do.
So what we need to do is we need to use these tools to secure all of our systems.
Yeah, that's right.
And the we here is the United States government needs to do that with its own systems.
The banks need to do that.
By the way, the tech companies need to do that.
By the way, this needs to be, individual consumers, individual people shouldn't be expecting to deal with this, but all of the tools that you have, the computer sitting in front of you right now, AI needs to be used to make sure that that's secure so that people can't break into it.
So every system from the most important military government system all the way down to the computer on your desk, these advanced AI should be used to secure these systems.
The tension is the AI that can be used to secure the systems can also be used to crack the systems.
Who gets access to the thing that can both secure and crack is like the hot government topic of the moment.
And specifically, that's what's in all the headlines in the last two weeks.
But it also makes me think of now, like my AI would be like a German shepherd in my house.
Even when I'm not here, it's watching and it barks.
And also when it knows how to attack.
I'm sure if it's programmed a certain way, if Mark comes over my house, it'll lick your hand.
But if it's someone who's an aggressor, it'll know how to distinguish.
And that'd be pretty easy, I think.
Obviously, there's going to be the camp of German Shepherds, and it's going to be escalation.
But the point is, if you just get rid of the dog and you leave your door wide open, how do you think it's going to end for you?
Here's my other big concern with AI.
So this has always been a concern about technology.
Oh, if the phone operators are out of work, they're going to be homeless.
Oh, you know, who's going to pick the cotton?
All this, whenever any invention occurs, people are hand-wringing that it's going to be the end of society.
That's never happened.
But, but, are we at a point now where the average human is like a horse?
Meaning an outdated mode of technology.
I'm thinking specifically of that like 50-year-old woman with no high school diploma.
She does ride sharing weekends to make some extra money.
You're not going to put her in the mines.
If the car is driving itself, if the algorithm is more personable than her and, you know, more likable than her, what role would you have for her?
Is it the case now that she's become outdated?
Right.
Yeah, so as you know, like this is an old argument.
By the way, Thomas Sowell also wrote about this, has written about this at length.
And so this has been a concern literally from the very beginning of the Industrial Revolution.
Right.
And actually, literally horses was like part of how this whole thing started, which was like in the beginning, like in the beginning, humanity, basically 99.99% of people were farming and specifically were farming by hand.
Yeah.
Right.
And then it's like, okay, if people aren't farming and you start replacing people actually with horses and with plows, and then you mechanize the plows and you mechanize the horses and you all of a sudden have like industrial agriculture, you know, and literally what happened over the course of 200 years.
was 99% of humanity went from farming to something like 3% of humanity went to farming, right?
Like 97% or something, you know, had to figure out something else to do.
And then by the way, what happened was food production went through the roof, right?
And food went from like super expensive and by the way, not very good to, you know, to just like, you know, keeping an abundance.
And by the way, you know, the great, you know, public health problem.
you know, used to be starvation and now it's obesity, right?
Even as you crash the number of people actually working in agriculture.
So yeah, that's the original version of this story.
That story has repeated itself a thousand times.
It repeated itself with, you know, everything, you know, railroads, it repeated itself with cars, it repeated itself, by the way, with computers.
There was a whole automation panic in the 1960s.
The magazines and the newspapers were obsessed with it at the time.
The computer brain was going to replace everything.
It's this thing.
And it's this thing where you have to basically kind of say there's this basically gap between, conceptual gap between, there's the jobs that we know about that are, quote unquote, at risk of being replaced.
And of course, there is some of that.
And then there's the creation side, which is like, okay, with all of the new wealth that's being created.
and all the new money that people have to spend and all the new interests and, you know, needs and desires and aspirations that people have that they couldn't even imagined.
You know, their ancestors 200 years ago couldn't have been imagined.
You know, 50 years ago, exactly.
And so one of the things that people can do on this to make it interesting, you know, you can do this with an AI now, but, you know, there's a U.S.
government department called the Bureau of Labor Statistics, and they actually track all the job categories in the U.S.
And you can go on their website and you can pull up all the job categories.
And it's a really mind-expanding thing to do because we employ people to do things today that our ancestors could have never even conceivably imagined.
I mean, Milton Friedman had a thought experiment on this once when he said, look, it's like, he said, human wants and needs are infinite.
You can never predict what they're going to be because humans are like relentlessly aspirational in the things that they want and need.
And by the way, the things that our starters want become needs, you know, very quickly.
Of course.
He said, you don't know what they're going to be.
You need to let the free market basically operate so people can basically find their own way and discover what they want and other people can figure out how to satisfy that.
And he said, look, you just need to be very open to all kinds of outcomes here.
He said, for example, the idea of the job of a therapist, right?
Like you pay somebody to listen to you.
right, would have struck your ancestors as completely insane.
You know, and then today it's something that like only wealthy people have access to.
And, you know, he's like, look, like in some future reality, maybe half the planet, you know, consists of being therapists for the other half.
Like maybe that's the job, right?
And he wasn't making a specific prediction, but he's just saying like, look, there's an aperture here for the creation of all kinds of new, you know, professions and occupations in response to this creation of nuance and needs.
I also think we have a particularly blinkered view of this right now because we've been living in a slow growth environment economically for our whole lives.
So one of the things that really happened, you know, basically since the 1970s is if you look at the history of economic growth in the West, economic growth used to be much more rapid.
Technological advances translated into the economy much more rapidly and the economy grew much faster, like as much as three times faster historically than it's been growing for our entire lives.
And so we've been living in a sort of a zero sum, a slow growth, zero sum, basically increasingly, as you know, like regulating bureaucratized environment with like less and less creativity that's translated into economic change, economic growth.
Like we think we've been living through an area of rapid technological change.
Economically, we've been living through a period of very slow change.
As a consequence, so much of our politics and so much of our psychology feels zero sum, right?
Where if one thing goes away, somebody else has taken it.
And this is, by the way, this is why you get political populism on both sides of the aisle is this kind of zero-sum fear.
AI is the first technology in decades that has the potential to dramatically increase what economists call the rate of productivity growth, which is basically the ability for the economy to grow much faster.
If that works and happens, then economic growth accelerates, right?
Then the economy starts.
So you just want to imagine the economy growing two or three or four or five times faster than it has historically.
And then as a consequence of that, all of this new discretionary spending money, you know, that comes out of people's wallets where they get to say, wow, I can collect art for the first time in my life.
And I really want to try this new, you know, whatever.
I, you know, would love to have a self-driving car and, you know, dot, dot, dot, dot, dot.
And people discover all these things that they want, that they can pay for.
And then all of a sudden you have this massive engine of job creation right behind that from all the people who are fulfilling all those needs.
Like, you know, I think there's a positive story.
It does require you to have, not even just optimism, it requires you to have an openness to creativity of the wants and needs that we don't yet understand in the industries that get built to fulfill those.
But what I'm saying is I don't see how that...
So this is the thing.
So people look at AI as the negative driver on this, and there will be some of that.
I mean, there is some of that where there will be certain jobs that are no longer required because the AI is doing it.
But AI is also superpowers to every individual person to be able to do whatever they want.
Right.
And by the way, maybe you can say this, this is like the massive split in the sort of the discourse of when people talk about AI and the abstract, they have all these like basically fears and anxieties.
Over a billion people are using AI already today.
Like more than a billion people use ChatGPT.
And the things that using ChatGPT for are the things that matter in their individual lives.
And they love it.
And it's great.
And everybody's, it's like huge numbers of people are using this to be better at work.
today.
They're using it to be better at work.
They're using it to learn new skills.
They're using it to do a better job, to make their boss happier, to be able to get promoted faster.
They're using it to start new companies, offer new services.
I mean, if you want to start a small business today, or by the way, you become a writer or like anything that you want to do, like the AI is like the best possible teacher.
coach, mentor that you've ever had.
It will like walk you through everything.
It'll teach you how to do marketing.
It'll teach you how to do sales.
It'll like, like, it's just like the level of capability that is being unlocked for ordinary people to have a, like a level of productivity in their life and in their work that they've never had access to before.
is amazing.
And so then all of a sudden, yeah, you get that person who was an Uber driver and now all of a sudden it turns out, it's just like, oh, wow, okay, I don't know.
All of a sudden there's like all these new tourists coming to my town and instead of driving them around, I can give them tours.
Okay, what would be the tour that they would like to go on?
Oh, well, the AI will help me like design the tour.
Oh, well, how do I start like a tour guide company?
Oh, well, here's how to do it.
Here's how you register it.
Oh, you know, how do I keep the books for a tour guide company?
Here's how to do it.
And the next thing you know, she's on the other side of that and she's in a completely different business and people are, you know, the Waymo car is delivering the tour participants to her, delighted to see her because they want a person to take them on the tour.
So that's the creative side of it.
Historically, the creative sort of generation of new ideas, generation of new wants and needs and the new businesses and professions that fulfill those wants and needs has raced way ahead.
right, of the replacement phenomenon.
You know, after 300 years of mechanization and computers, there are more jobs in the world today and at higher incomes than ever before in human history.
I think that's exactly what's going to happen here.
And I think basically people who, I understand the concern, but I think it's just, I think it's fundamentally a failure of imagination.
And I think that the human spirit is going to process this just fine.
I think that where you and I disagree, or maybe I'm not understanding correctly, is I think a huge significant...
segment of the population, let's say a third, I'm being conservative in my opinion, are not capable of self-direction.
So if someone is, that woman who's the tour guide, that's an easy one because that's someone who's like, okay, what should I be doing?
Okay, I'll learn these skills.
It'll take me a day.
I'll do my reading and I'm personable and they're going to clean up.
That's an easy one.
But I think there's plenty of people who are basically just want, Mankin had said, the average man does not want to be free.
He merely wants to be safe.
There are people who just are wired that they want to be told what to do.
And at a certain point, I don't see what value they're adding to any company or anyone else.
And what do you do with those people?
Just put them in welfare.
Well, I mean, this gets an area of social policy and, you know, political theory that maybe I may stay away from.
at least live on video on the internet.
Okay, sure.
But I guess I would say this, like one of the things you can talk to AI about is this problem.
Like one of the things you can do is you can say, wow, like I don't know what I'm going to do.
Like here's what I've done for my entire life.
And wow, I don't know, it seems like my job's going to get like replaced.
It's like, okay, what should I do?
The thing will give you career advice.
Right, like it will tell you, like if you want it to, it will tell you what to do, right?
And I don't think people should just like ask it what to do and just do what it says, but you can say, okay, brainstorm with me.
I use it for brainstorming.
This is an area you can use for brainstorming and say, all right, look, like here's what's happening.
Okay, well, it's gonna say, well, where do you live?
What's going on?
Here are the different areas.
What are the trends?
What are the new things that are happening?
You know, I don't know.
I, you know, every time I go by my local gym, I see there's like six new kinds of exercise classes I never thought of.
Could I become a, you know, could I become a personal trainer?
You know, okay, what's the hot new trend?
How do I get, you know, certified on that?
And then the next thing you know, you're doing that.
And the thing will just like happily, it'll, it'll, it'll.
So one of the amazing things about this, I think this is really underrated.
These things, the AI models, it's like the best doctor you've ever had in your entire life.
Like it's amazing being a doctor.
Like it will hold your hand through any medical situation you're in with a degree of like caring that people are not, like beyond what a human doctor, not only what most human doctors will just do naturally, but no human doctor has time to do it in the way people really need.
It's also the best lawyer you've ever had, right?
It's also the best like coach you've ever had.
It's also the best ghostwriter you've ever had.
It's the best editor you've ever had.
And it's the best advisor you've ever had, right?
And so if you give it that opportunity, and again, you don't have to answer, you can actually ask the questions.
It will answer the questions.
Yeah, what should I do?
How should I think about the evolution of my career?
How should I think about the change in the economy?
It will happily do that with you.
And in a way that people literally have never been able to have that conversation with people before.
You know, you really kind of solve that question in my mind because I have a good friend and he's made a lot of money in crypto, but he's been sitting at home not having to work and it's been driving him crazy.
So he's like, I need to have a job.
I don't need to make money.
I just need, I'm like, make candles for guys, scented candles for men.
Like that's like, it's, you can go down that rabbit hole, work at the sense, blah, blah, blah.
It'll occupy your brain.
You'll create a product, even if you make $50, who cares?
And what I'm realizing is if you ask AI, it will have that Venn diagram of no one has made socks for 14-year-old immigrants.
It'll see those where there's holes in the market and then it can walk you through it.
So it really does, except for the people who completely bring nothing to the table, which that's fine, that's a whole separate conversation.
But that woman with the, you just really solved my question because she can be like, okay, what value can I bring?
What could I produce?
Make cookies.
There's so many things that people always want that personal touch.
And also there's so many, the markets will get more and more niche and AI will be perfectly able to find, no one is talking to people who like saltwater aquariums, but also like heavy metal music.
So start a heavy metal saltwater aquarium website.
There you go.
Holy crap, you just answered my question.
I would love a heavy metal aquarium.
It's like Angry Little Fish.
Yeah.
Like, why not?
A hundred percent.
Yeah, no, totally.
Well, here's maybe another way to think about it.
Oh, I think that's right.
Here's another way to think about it, which is, you know, people think about this as like, well, if the machine does something, it's dehumanizing.
I don't know about you, but like office jobs are dehumanizing.
Like, oh yeah.
Like sitting in a cubicle for eight hours a day, like working on the same thing over and over again.
And then by the way, factory jobs are dehumanizing.
And by the way, like I worked, you know, I grew up in agriculture country.
I worked on farms.
Like, you know, farming is not romantic.
I'd be, you know, people sometimes.
City people sometimes talk about their romance of going off and doing like whatever boutique, you know, farming.
And it's like, you know, how do you feel like getting up at six in the morning, going to bed at midnight, working like seven days a week for the rest of your life?
It's Sisyphus.
Yeah, like it's like, yeah, exactly.
Like you're in, you're fighting back like chaos, you know, the entire time.
And so like, I think just so much of like what, it's almost like we have collective PTSD.
Like we all have the, we've all had to live these lives with a level of, like, if we look at our ancestors and the lives I lived 300 years ago, we just are like, that was a level of drudgery and poverty and limited options that would drive us crazy and make us want to kill ourselves.
Our ancestors 30 years, I mean, 300 years from now, even 30 years from now are going to look back at us being like, I cannot believe they did that.
I cannot believe they spent time doing those things.
Like that was such a waste of human potential.
That was such a waste of human creativity.
And then maybe one more.
There's one more thing to kind of add on that is I think that this goes to the thing of like, does technology like this make us less human or more human?
And I think like the examples we've been using, I think you start to see why this can result in people being more human, which is if the physical needs of data in their life are more easily satisfied, then people can spend more time actually being human.
and then they can spend more time actually and money spending time on things that actually are human experiences.
By the way, another just incredible version of this is already playing out.
I got to interrupt you just to validate your point of view because back in the day, I'm reading a lot like the early socialists, 1890s, 1910s, people working 16 hours a day.
And even just down to 10, which is still a lot, then you're complaining people watching too much TV because they have too much spare time.
That's right.
Right, right.
All the diseases of scarcity become diseases of abundance.
And diseases of abundance are still diseases, but they're better.
Like obesity is better than starvation.
And then, by the way, then you figure out how to solve obesity later on.
Yeah, I mean, look, I'll just give you a micro, one example of this was happening in music, right?
And so, you know, once upon a time, all music was in person.
Like the only time you'd ever hear music was like if you happen to stumble into a church or something and hear it for the first time.
And by the way, maybe the only time in your life, you know, there were people who maybe heard music once, you know, in their entire life, right?
And so, and then, and then, and then, by the way, by the way, then sheet music appeared.
Actually, there was a whole moral panic about sheet music when it first appeared because it was going to put all the, all the pianists out of business and they all got extremely upset.
But then that, you know, led to ultimately recorded music and then obviously digital music and streaming music.
And now, now if you talk to any musician, it's like, can you make money, you know, with recorded music?
It's like, no, you know, not really anymore.
You know, you can get your music distributed on Spotify, but you, you know, you hear the endless complaints about, you know, you get back pennies or something.
And so the recorded music business is not what it used to be.
Of course, what's exploded is live performance, right?
And so live music is like exploding through the roof.
And now you see, and now the complaint, right, is the tickets are too expensive to go to the concerts.
Well, it's like, well, okay.
But if you think about it for a second, it's like, why are we still listening to live music?
We all have every piece of music ever written available on demand in high fidelity in our homes, on our earphones anytime we want, essentially for free.
why is anybody going to a concert?
And of course, the answer is because the concert is a human experience, right?
And so, of course, when we get discretionary money, we want to go have the human experience.
We want to go have the concert.
By the way, if we're going to throw a party and we want to be a very special party, do we play music through speakers or do we hire musicians?
We hire musicians.
And also going to the concert, you go with someone, you create a bond, which is a hundred percent, a hundred percent.
And so there's another way to think about it is every profession that involves human to human contact is going to go bananas.
Right.
Right.
And, and, and I, and I think that those jobs, by the way, that's going to be great for all the people getting experience that.
And then I think those jobs fundamentally are better jobs, like just at a, at a, at a, at a very, at a very core level and they are going to go, it's going to be a bonanza.
For the first time in this show's long and sordid history, you have completely, I'm not kidding at all, you have completely answered all my concerns about the topic, and I'm...
I totally get it now.
Thank you.
I feel so excited.
And since you're much more visionary about this stuff than I am, because you've been there from the beginning, I can't even imagine how exciting it must be for you.
Things that people were kind of hypothesizing about even five years ago, now you're using them on a day-to-day basis.
You must be absolutely giddy.
So the book is The Techno-Optimist Manifesto.
You got it, Passage Press.
Mark, thank you so much for taking the time.
We're running out of time.
What has been your favorite part of this interview?
The amazingly, amazingly probing questions.
You are welcome.
Thanks again for listening, and I'll see you in the next episode.
but A16Z does not guarantee its accuracy.