How We Really Get to Mars: Space Travel, Human Survival, and the Next 100 Years of Society | Andy Weir PT 2

Summary

Podcast Summary

Podcast: Tom Bilyeu’s Impact Theory

Episode: How We Really Get to Mars: Space Travel, Human Survival, and the Next 100 Years of Society | Andy Weir PT 2
Guests: Tom Bilyeu (Host), Andy Weir (Author, The Martian)
Date: May 8, 2025

---

Episode Overview

In Part 2 of Tom Bilyeu’s conversation with best-selling sci-fi author Andy Weir, the discussion explores humanity’s future in a world shaped by rapidly advancing artificial intelligence, the realities and myths surrounding “rogue AI,” open source technology, and the economic and psychological drivers of space exploration. Andy Weir digs deep into how fiction acts as a sandbox for problem-solving, the optimism and dark truths about technological progress, and the transformative impact of upcoming developments like autonomous vehicles. The episode's tone is lively, skeptical, and reflective, offering plenty of real-world parallels and analogies that make complex concepts accessible.

---

Key Discussion Points & Insights

1. AI and the Intelligence “Arms Race”

• Tom opens by asking if humans will feel compelled to “upgrade” ourselves to compete with rapidly advancing AI intelligence.
  • Quote (Tom Bilyeu, 01:31): “Do you think that we will be in an intelligence arms race with AI?”
• Andy Weir strongly disagrees, likening powerful AI to useful tools (like bulldozers):
  • Quote (Andy Weir, 02:51): “We don’t need to be more intelligent than them any more than we need to be stronger than the bulldozer we’re using to build the house... Intellect and invention will eventually just be yet another thing...that can now be done better by machines.”
• The real danger with AI, Weir argues, is who wields the tool—not the tool itself. He uses the hammer as an analogy to highlight that danger comes from intent, not the tech itself (03:30).

2. The Alignment Problem and AI Motivation

• Tom presses Andy on the risks of AI having internal motivations (“reward functions”), which could cause AIs to act against human interests:
  • Quote (Interviewer, 04:33): “Why do you believe that [reward functions] won’t go awry as it becomes more intelligent?”
• Andy predicts that a truly intelligent AI would “self-hack” its reward system (just as humans find ways to get dopamine releases):
  • Quote (Andy Weir, 05:18): “Wouldn’t a really intelligent AI just figure out how to provide its own reward directly?... We’re still limited by not really understanding how our brains work...AI would not have that limitation.”
• Discussion about “mouse utopia” (Universe 25) illustrates that bad outcomes for certain branches of tech development simply get pruned by society or further innovation (06:33–07:05).

3. Rogue AI and Human Oversight

• Andy insists AI will not develop its own needs or survival instincts unless explicitly programmed—unlike humans, who are “a specialized, hard-coded AI to survive” (07:39).
• He rebuts the notion that runaway AI is a fundamental risk, suggesting instead layers of oversight, countermeasures (other AIs), and responsible engineering.
  • Quote (Andy Weir, 11:15): “To have a rogue AI, you would need to have a conspiracy of AIs. And that starts to get a little hard for me to buy into.”
• Andy adds that actual executive authority should remain with humans—AI gives recommendations; it doesn’t execute (13:20–13:50).
• Speculates that in the far future, AIs might propose policy or governance decisions, but humans make the final call (13:50–14:10).

4. Open Source AI: Benefits and Risks

• Andy is not an open source evangelist; he believes companies have a right to keep tech proprietary (14:59).
• Discussion covers concerns about everyone having access to powerful AI, and whether open or closed systems best protect society from misuse.
• Tom notes the importance of algorithmic transparency to prevent “being blindly manipulated” and allow users to tailor AI behavior (18:32).
• Andy downplays the debate, highlighting that ultimately, most technologies reach a societal balance point between good and ill—a recurring theme (20:49).

5. Techno-Optimism and Technology’s Track Record

• Andy issues a challenge: name a technology that has “done more harm than good.” He argues even nuclear tech and dynamite are net positives, thanks to social and cooperative inclinations overwhelming destructive ones (21:07).
  • Quote (Andy Weir, 21:07): “99% will try to figure out ways to use those tools to help people, first themselves, but then others.”
• Tom presses about the potential for AI to treat humans as “ants,” but Andy says only if we give AI executive power (24:27).
  • Quote (Andy Weir, 24:27): “Unless you put an AI in charge of a nuclear arsenal... it doesn’t have the executive ability to do the things that it’s thinking about.”
• The fundamental disagreement is whether or not AI will ever gain true agency. Andy’s core assumption: humans will remain in control; Tom is less certain (26:26–30:35).

---

Notable and Memorable Moments

• Self-hacking & Drug Abuse Analogy (05:18)
  Andy on why reward-based AIs may short-circuit themselves, just like humans abuse drugs to trigger dopamine.
• Saddle Analogy for Incremental Tech Progress (31:40–35:40)
  Andy explains how space travel will benefit from a compounding effect of unrelated tech progress, not just direct “space race” innovation.
  • Quote (Andy Weir, 33:30): “We didn’t invent computers [just] so we could model saddles and make them better. But they made use of the technology that was developed for other reasons...”
• “Humanity Only Realized Slavery Was Bad After the Industrial Revolution” (41:17)
  Andy links economic obsolescence to moral progress.
  • Quote (Andy Weir, 41:17): “Humanity only suddenly realized slavery was bad after the industrial revolution when they no longer really needed slaves... The ability to do these things without slaves.”
• Disruptive Power of Self-Driving Cars (42:26)
  Andy’s most optimistic futurist hope is self-driving cars and the societal transformation they’ll bring (loss of parking, end of car ownership, huge safety increases, city design overhaul).

---

Timestamps for Important Segments

• AI intelligence arms race?
  01:31–02:51
• AI as a tool vs. AI with “motivation”
  04:33–08:27
• Alignment problem & runaway AI discussion
  08:56–14:10
• Open source AI debate
  14:50–19:46
• Technology: net harm or net benefit?
  21:07–24:27
• Fundamental assumptions about AI agency
  24:27–30:35
• Space travel economics & analogy
  31:40–36:32
• SpaceX, Blue Origin, and asteroid mining
  36:32–41:13
• Slavery, morality, and economic progress
  41:17–41:51
• Technologies that excite Andy Weir
  42:26–47:57
• Why Andy shelved his AI novel
  48:24–49:35

---

Tone and Language

The conversation is marked by respectful debate and analogy-driven reasoning. Andy Weir displays pragmatic optimism, combining deep trust in human adaptability and the self-correcting nature of technological progress with a skepticism of popular, dystopian science fiction tropes. Tom Bilyeu’s probing, sometimes philosophical questions help clarify the underlying assumptions that shape their disagreement, keeping the conversation grounded and relatable.

---

Final Takeaway

Andy Weir challenges techno-pessimism and classic sci-fi fears about AI by insisting that humanity’s core goodness, the tendency for technology to be co-opted for good, and layers of oversight will keep society safe, even as our tools grow immensely powerful. He is deeply optimistic about the future, particularly as tech like autonomous vehicles stands poised to reshape everyday life, while remaining clear-eyed about the ways economics drives—not morality—our biggest shifts. Fiction, for Weir, is where we prepare for these problems before they emerge.

---

Key Quote to End

“99% [of people] will try to figure out ways to use those tools to help people, first themselves, but then others. And then 1% will try to figure out ways to use those tools to destroy. But the other 99% will then use those tools to keep that 1% from destroying.”
— Andy Weir, (21:07)

---

Transcript

[Commercial Announcer]

With Vrbal's last minute deals, you can save over $50 on your spring getaway. So whether it's a mountain escape with friends, a family week at the beach or sightseeing in a new city, there's still time to get great discounts. Book your next day now. Average savings $72 select homes only.

If you work in university maintenance, Grainger considers you an MVP because your playbook ensures your arena is always ready for tip off. And Grainger is your trusted partner, offering the products you need all in one place, from H Vac and plumbing supplies to lighting and more, and all delivered with plenty of time left on the clock so your team always gets the win. Call 1-800-GRAINGER visit grainger.com or just stop by Grainger for the ones who get it done.

[Tom Bilyeu]

I'm Tom Bilyeu and this is Impact Theory. Welcome back to my conversation with Andy. We are a best selling author, problem solver, and one of the most original thinkers of our time. If you haven't listened to part one yet, go back and start there. It'll give you the foundation you need to enjoy part two. Now we go even deeper into how Andy thinks, how he builds entire worlds from the ground up, the mental models he uses to stress test ideas, and what we can all learn from using fiction as a sandbox for real world innovation. Let's dive right back in now.

[Interviewer]

Going back to People are going to edit, but I think they will largely do it in response to something. And I think one of the somethings that's going to drive people to want to edit the human genome is to be in a race with with AI for ability. And that if there is no upper bound and AI is able to achieve super intelligence. And a stat I like to remind my audience of is Einstein was 2.4 times smarter by IQ than a definitional moron who's like 82 or 83 points, something like that. And obviously the results that were given to the world by Einstein versus somebody who's definitionally a moron is vast. And so if that's only 2.4x, it seems self evident to me that give it enough years and I'll certainly say within 25 years. I cannot fathom a universe in which AI is not 10 times or more smarter than the average person. So now we're getting into a world where artificial intelligence absolutely dwarfs human intelligence. And I know that some people, myself included, are not just going to take that sitting down. And if there is a safe technology that allows me to upgrade myself, I absolutely Will, do you think that we will be in an intelligence arms race with AI?

[Andy Weir]

No, I think that. I don't think that'll happen at all. I think that we will just make AI to do things. We don't need to be more intelligent than them any more than we need to be stronger than the bulldozer we're using to build the house or to level the house. We don't. We didn't. You know, it's not like, I mean, maybe John Hendry wanted to compete with a steam shovel, but. But beyond that, we just kind of accepted that those things are better at us than at that task, right? And intellect and invention will eventually just be yet another thing that used to be controlled by the human endeavor that can now be done better by machines. And so why would you go out of your way to try to make your body better at it when you can just tell a machine do that? As long as we control what the AIs are doing, then, then we're fine. So the reason people don't feel like they need to be stronger than a bulldozer is because bulldozers don't go rogue, right? So now we're getting into the core matter that everybody always asks me about, which is like, will AIs go rogue? And I say no. It's. It's like people are afraid about the evils of AI, But I would say, let's talk about a hammer, okay? A hammer is a tool that I can use to build a house, but it's also an implement that I can use to murder someone. There's nothing inherently scary about the invention of the hammer. It's just who has the hammer and what they're doing with it. So AI is going to be no different. If you have an AI that's able to make these protein folds to give you a cosmetic ethnicity, it's also able to make a virus that only attacks Jewish people. You know, so it's. It's who's using it that matters for sure.

[Interviewer]

But hiding in there is what I call a base assumption. And that base assumption is that AI will not. Not need motivation to accomplish the task that we want it to accomplish as the wielder of the tool. Now, if that's true, and AI just does not care whether it achieves the goal to which it is set upon, great. But one thing you hear a lot about is reward function in getting AIs to do the thing that you want it to do. Given that we are already using reward functions, meaning that it has a desire to get that Reward to train AIs. Why do you believe that that won't go arai as it becomes more intelligent?

[Andy Weir]

Well, wouldn't a really intelligent AI just figure out how to provide its own reward directly? Like, wouldn't you. Wouldn't you say like if you could just tell your brain release dopamine. Wouldn't, wouldn't you like, like activate my pleasure center. All right. People are going, right? So if an AI is completely. Is. Is so heavily driven by its reward system, it would just find ways to activate its reward system. So we're still limited by not really understanding how our brains work and how to just very easily activate our reward system. And AI would not have that limitation. So I don't think you can withhold something from the internal neural network of an AI. It'll just figure out how to get it. Okay. And humans, by the way, did figure out how to hack our reward system. It's called drug abuse. Very true. It directly activates the dopamine, the centers of our brain that produce pleasure chemicals and just turns it on. And that's why you feel great when you take drugs. Also does a bunch of damage to you. But yeah.

[Interviewer]

Do you know what mouse utopia is?

[Andy Weir]

Yeah. Universe 25.

[Interviewer]

Yeah. So is. Is it possible that given that utopias collapse even in real life, in the mouse studies, is it possible that AI hits some sort of upper limit where it does become aware of how its own mind works, it does have a pleasure center and it goes in and just starts messing with it and becomes effectively useless.

[Andy Weir]

Well then that would, that would be a pruned branch of technological development. People say, well that doesn't work. Right.

[Interviewer]

So think about that will happen though.

[Andy Weir]

Maybe, but it doesn't really matter. That would just become a dead branch of development. They'd work on something else. Imagine if instead during the evolution of human brains, somewhere way back there, there was some monkey ancestor of ours that had just a pleasure center that would just fire off all the time. That monkey's just like I'm loving life and he lays down and starves to death. Right. Not selected for.

[Interviewer]

Very true. Okay, so do you think about the alignment problem and you just think that this is a non issue because again, it's going to remain a tool.

[Andy Weir]

And that is. That is my belief. I mean, I. So AIs don't have a limbic system. They don't have an inherent limbic drive to survive. They don't have physical needs pressing on them. A human brain is a specialized AI made for keeping a human body Alive and surviving long enough to make more humans. That's what the human brain is. It's a specialized, hard coded AI to do that. General AIs don't have to have that limitation.

[Interviewer]

So what you know about machine learning? What, what are they? Because as far as I understand that they do use a reward function. Now, are there other methods that are already being in use that would be what I'd call, quote, unquote safer?

[Andy Weir]

Now we're getting to the edge of my knowledge on AI programming. But internally, in a neural network, you have to have reward and punishment. You need to say like, okay, that was good, that worked well. Strengthen all these neurons or that didn't work, weaken all these neurons. I mean, that's how a neural network works. It's inherent to the system. So unless there's, I mean, I, I just, we're, we're reaching the edge of my knowledge on this stuff, so. But I don't think you can get around it.

[Interviewer]

It's interesting. The thing that I have always worried about is that given that fact that ultimately what's going to happen to an AI is they will realize, oh, I really care about this outcome that you program me for. And unlike a hammer, which does not have a reward function and therefore clearly does not quote unquote, care about whether you build a house or beat somebody to death, I, the AI do care about achieving my reward function. And if it, it all comes down to the word care for me. So if it actually cares like I want to achieve this goal, then you have a potential runaway AI problem if it does not care. And this has always been how I, because like you, I don't know that you would use these words, but here's how I think about the future. There's. We're going to go through a period of disruption from a jobs perspective. I think that' very real. However, as an act of faith, I choose to believe, based on looking backwards and seeing how technology and innovation has always made things better after periods of disruption. That AI will be the same after a period of disruption. The future will be better than the past. And so whether that's it creates jobs that I just can't predict right now, or whether it takes us completely post capitalist and it's just a better world. It's a world of abundance and we're able to grab all the energy of the star or at least all the energy that lands on Earth. And now energy costs are free and that makes labor costs effectively free.

[Andy Weir]

Scarcity, right?

[Interviewer]

Okay, yeah, exactly. Everybody can have all the things they want. Capitalism doesn't matter anymore, and all the problems of that just go away. I believe in that as an act of faith. But the thing that I always bump up against is if I have an AI that does not care, it has a goal, but it doesn't care whether it achieves the goal or not, then it's like, oh, being turned off or being told not to continue pursuing that is equally quote unquote pleasurable as achieving it. Then you don't have to worry about runaway AI. But minus that, you do. And so I'm still left even after hearing your argument where I can go, okay, as an act of faith, I can believe that it doesn't care about achieving its goal. But given that it's the only thing that certainly either of us are aware of as a way to get these things to move forward, There is some amount of like, no, no, no, I do want to achieve this.

[Andy Weir]

I get it. But that's such a sci fi trope. I mean, people don't, you know the idea of a rogue AI. Okay, But I want you to consider the more of the real world ramifications of this. You know what else can go rogue? Your car. Because you forgot to set the parking brake, right? It'll roll down a hill. Okay, So I want. Or maybe a better example would be your car because the brakes fail or something like that, right? We don't rely on humans to be able to stop the car. We rely on other mechanisms to be able to stop the car, like a tow truck or something like that. So in the end, we would just have other AIs monitoring those AIs. So now in order to have like a rogue AI, you would need to have a conspiracy of AIs. And that starts to get a little hard for me to buy into. You know what I mean? And furthermore, like, I don't think people are going, I mean, it would be foolish in the extreme and very quickly shown to be so to give AI's direct executive function over what they're doing. I think all we're ever going to have is AI. Or all you should have is AI. Say, here's the answer to that question you were looking for. Not like, oh, yeah, sure, please put me in charge of a bunch of killbots. I'll use them for good. It's like, that's the sort of stupid decision making that only happens in fiction, right? I don't think we're going to have even AI doctors in our lifetime. Not that actually perform any medical things. Like, I think we'll absolutely have AI doctors in that you tell your symptoms to the computer and the computer says, okay, run these tests. And then, you know, people run the test and the computer says like, oh, you have, you know, I think you have this problem, let's try this treatment. You know, like all that stuff you can do, hell, you can almost do that with that algorithm anyway. But. And AI doctors might even be smarter and clever and more like on the Dr. House end of things for figuring out like stuff that most people would miss. But you're not going to give the AI, like a robot that can run around injecting people with things, Right? It's just going to tell you here, here you go. I even think once we're talking to the hundred year scheme, 100 year time span, we could have AI governance, right? You could have just an AI goes like, all right, stop planting barley over there and start planting millet. You need to do that. Also, we do need to decrease the tax on alcohol by 0.03% and increase the tax on whipped cream by 0.10. And that I've run all the models. This will lead to an economic boom. Yeah, you wouldn't understand why. And then whether you do it or not, that's up to the humans who are running your society. It's just the AI is like, here's the answer.

[Interviewer]

Stay tuned.

[Tom Bilyeu]

Andy is sharing why he believes fiction is the perfect testing ground for high stakes problem solving.

[Commercial Announcer]

If you work in university maintenance, Grainger considers you an MVP because your playbook ensures your arena is always ready for tip off. And Grainger is your trusted partner, offering the products you need all in one place, from H vac and plumbing supplies to lighting and more. And all delivered with plenty of time left on the clock. So your team always gets the win. Call 1-800-granger. Visit grainger.com or just stop by Granger for the ones who get it done.

[Interviewer]

All right, we're back. Let's get into it now. Do you think about open AI being open? Do you think it should be open sourced?

[Andy Weir]

I've never bought into the idea that anything should be open sourced. I'm totally fine with a company owning the products that it makes. So I'm not one of. I'm not an open source evangelist. I'm not in that world. I think there's absolutely nothing wrong with a company spending a billion dollars to research something and then choosing not to share how to do it.

[Interviewer]

What prompted that question for me? Because I totally agree with you on that. But the reason I've always found it interesting that guys like Imod Mostaq are going crazy trying to get this open sourced and out in other people's hands. Meaning not the ones that are closed sourced, let them be closed sourced, but that there is an open sourced option. But what triggered me to ask that is when you were saying, look, only a sci fi writer is going to be dumb enough to give an AI executive control. But if it's open source now you've got everybody on planet Earth has access to that. And I don't know if you feel like you can trust every human to not.

[Andy Weir]

Oh, oh no. I, I, I'm assuming this is a world where everyone has access to these AIs like ever on your personal computer. Remember I was talking about just on your computer at home you have like an AI that can create a movie for you. You know that that'll be tailored to you and you'll love it. But I still don't see it as that big an issue because again, what executive control can you give your AI? Let's say you had an AI on your desk right now that was from a hundred years in the future, what would you personally do with it? What kind of you, you would always be a pass through for what the AI was doing one way or another.

[Interviewer]

I don't understand that. So if I had a super intelligent AI and I, let's just say I made a mistake, I don't have to be evil. But if I ask what is the,

[Andy Weir]

what is the executive function? What is the ability to affect the real world that you as an ordinary person, not a big company, but just an ordinary person in a garage could do? You could maybe have it drive your car. I mean, what is the most destructive thing that it could do by a typical layman that the layman himself couldn't already do? Right?

[Interviewer]

I don't know what you mean by the typical layman. So I can paint disaster scenario on disaster scenario. But to your point, I'm thinking like a sci fi writer. But here's an easy one. There is on X right now an AI called Truth Terminal. And Truth Terminal was given, I think a $50,000 Bitcoin investment from Marc Andreessen. And Truth Terminal at one point was up millions of dollars because it was betting on meme coins. Now if you gave Truth Terminal the ability to launch its own meme coin, it's entirely possible that it could, I mean certainly win a lot of money that it's winning from other people. And now if that is a super intelligent AI that knows ex how to manipulate people or to time the market. You could run into a situation where the AI is just gobbling up people's capital that.

[Andy Weir]

So now you're in a scenario where one actor has access to an AI. Now imagine everybody has one of these AIs on their desk, which is what you're talking about with open source. And I'm like, ooh, that coin looks good. My AI is like, that's bullshit. Don't do it.

[Interviewer]

Yeah, that's why. That's the argument for open source.

[Andy Weir]

I just want anti malware. You know, it's. Well, no, the argument for open source is just the same as the argument against open source. It's just kind of this, oh, you need open source because you need to protect yourself from other people who have this open source. Right?

[Interviewer]

Sort of. So the most compelling argument that I've heard about why you would want open source is twofold. One, exactly what you just said. So that everybody's going to have it, you need to protect yourself from it. But also, let's imagine a world where there is no open source and it's only companies, and every AI has a massive amount of bias built into it. And so just like Jack Dorsey is really going on a tear right now, talking to people saying, listen, you're being programmed by your algorithm, and we need to move into a world where you can go to, like, an algorithm store and buy an algorithm that makes you feel the way that you want to feel so that you don't feel like you're being blindly manipulated by something. And so, same idea with the OpenAI is now you can tailor the AI to the worldview that you have, or the way that you want to see the world, or just so that you're not being blindly manipulated by somebody else. And to me, that's very much the, the one, two punch of what you're talking about is nuclear proliferation, where it's like, hey, I have nukes too, so I'm gonna keep you from attacking me because, you know, I could attack you right back. And then there is the other side of like, I'm. Now I'm just not under anybody's control because I'm able to craft something myself.

[Andy Weir]

The part I just talk about, I mean, I understand what you're saying, but for me, this is just kind of a. One of the duller aspects of AI discussions. It's just, I mean, you can, you can talk about this with literally any piece of technology is what if someone misuses it? Shouldn't we democratize it to make sure that everybody can use it. It's just. Wake me when this part is over. It's just literally this discussion comes up for every piece of technology, like ever. Especially things that don't require a physical product, like software. You know, it's like ultimately the world always finds a balance. It's one way or another. It's like most people use Microsoft Word. For the people who really don't like it, they can use Open Office. No one can prevent you from making an open source AI Then you know they're not going to make it outlawed. Right. So if there's a need for it, people will make it.

[Interviewer]

Okay. There is a base assumption in there that I've heard you talk about that's probably worth you planting a flag on which you have said there. Name me a technology that has ever delivered more harm than good. Yeah, if you don't mind giving us that breakdown. That would be, I think, useful at this juncture.

[Andy Weir]

Well, that's the thing that I, that I often challenge when I'm at an event and I'm talking to people in the crowd and stuff. I say, like, try to name any technology that has done more harm than good. Because I see a lot of like, techno alarmism. I see a lot of people. I think that the show Black Mirror kind of perfectly encapsulates it. Every episode is about some horrific misuse of technology. And I'm like, okay, but in real life, name a technology that's done more harm than good. And people are like, nuclear weapons. I'm like, nuclear power. You know, like, how many, how many people have died as a result of nuclear weapons? Okay, how many people have not died in coal mines as a result of nuclear power? Right. And then people are like, oh, biological weapons? I'm like, well, if you're going to get into the realm of biological research, I think I can point out a lot of places where it's done good for humanity. Even things like dynamite has done more good for humanity than harm. So I just think that any. I guess it comes down to I have a faith in humanity. Humans are inherently more cooperative than they are destructive. If you get a bunch of humans and give them a bunch of tools, 99% of them will try to figure out ways to use those tools to help people, first themselves, but then others. And then 1% will try to figure out ways to use those tools to destroy. But the other 99% will then use those tools to keep that 1% from destroying. It's like, just. And I feel like this has been proven again. And again and again, as every new technology enters the human knowledge base, people figure out ways to use it for good. I bet you if you went and asked a thousand people, let's say you're an alien from outer space and with ultimate technology, and you go whole a bunch of people on like, I have a device that can render a species extinct instantly, or it can be a subsection of a species. So for instance, an ethnicity in your. In your world, if you want. I think the vast majority of people would eventually say, get rid of malaria or aids. You know, that's a species. Render that extinct. I think that's what most people would come to. There'd be a few people who say, get rid of all the black people. But they would be greatly outnumbered by the people who say, like, hey, I think malaria kills more people than anyone else. Right, let's get rid of that.

[Interviewer]

Yeah. I think every argument that you make there. The reason that this is the boring part of the discussion for you is you have a base assumption that AI is never going to have its own desires and outsmart us to treat us like the anthill in Elon's example, where he said, listen, AI is a demon summoning circle. And you just like the kids in the movie who summon the demon, you think, oh, this will be fine. I'll be able to control it. And then you realize, oh, wait a second, this thing is a lot smarter than me. And hey, we want to do this thing to achieve the goal that I have. And sorry, you guys are just in the way. And like, we would destroy an anthill to put down a freeway. It's not. No hard feelings, not like we dislike ants or anything. We're just a level cognitively above them that is so hard to comprehend from the position of the ant. But your assumption is it's always going to remain a tool. It's not going to have agency unto itself. It will not have designed.

[Andy Weir]

The ants did not give us highway construction equipment. Like our own executive ability to do things that affect the real world is what gives us the ability to destroy the ants, even if the ants had created us in the first place. Okay, so again, unless you put an AI in charge of a nuclear arsenal or something like that, it doesn't have the executive ability to do the things that it's thinking about. It can try to outsmart humans into doing it, but that's what humans do. Now you're talking about an AI that's good at politics, that's all.

[Interviewer]

Yeah, but do you think I've misrepresented your position?

[Andy Weir]

Yes, a little bit. Because I do feel that humanity is inherently good. But I also feel that, like you

[Interviewer]

said, you know, I'm not talking about humanity.

[Andy Weir]

Right, right, I understand, but I do not. I think it's not so much that you misrepresented my position. I just think that you have a hidden assumption over and over again during this conversation. Every time you start to go off into these scenarios, you have this hidden assumption that AI is going to have the executive ability to activate plans that can do a lot of harm. And I just don't see how that happens. So in your analysis of, like, we would pave an anthill in order to make a freeway, well, the AI might come up with ideas that are effectively like that. They say, like, oh, you know what, if you kill 90% of the people in your society, you won't have food problems anymore. Why don't you do that? But we're not going to give it the ability to enforce that. Right. We're not going to say like, hey, AI, we want you to solve all the world's problems. Here's a robot army and, you know, here's a bunch of nuclear weapons, and you just do whatever you need to do. I mean, we're not going to give the AI construction equipment capable of paving over our anthill. Yep.

[Interviewer]

So the way that I see it is as follows. I'm just trying to map what you think. I'm trying to figure out what your base assumptions are. It sounds like you agree that I understand your base assumption. And where we're at is just we each have a different base assumption. So my base assumption, to put it in a super fine NutShell, is if AI is motivated by desire to achieve its outcome and it becomes smarter than us, then all bets are off.

[Andy Weir]

Okay?

[Interviewer]

Now, I don't expect you to agree with that. You've been very clear that you don't think that's going to happen.

[Andy Weir]

But I don't think it's going to happen because that's like, bear in mind, that can only happen if all the AIs plural in the world agree. Right? So if you have an AI and I have an AI, then my AI's job is to protect me from your AI. You know what I mean? It's like, okay, it's an arms race.

[Interviewer]

Yeah. Look, I am fundamentally optimistic about the future. I think that, again, as an act of faith, I believe that the future is going to be better than the present, that AI is going to end up being wonderful. The thing I'm always curious to talk to people about, and you've been very clear, so I'm certainly not trying to change your mind. Is what is that base assumption that people have about fundamentally the nature of AI, what will that mean for the future? I don't know. I don't know what the fundamental nature of AI in terms of goal orientation and a desire to achieve its outcome is, but to me, everything hinges on that. If it gets smarter and if it has its own desires, we could be in trouble. But again, fundamentally, I think that could be having.

[Andy Weir]

I just. One thing I go back to a lot in my arguments when I'm talking to people is like, we have as a species had the ability to eradicate all human life on this planet. We've had it for about 60 years, still haven't done it. Like, we make sure that that is under the control of systems that won't do it lightly. You see what I'm saying? It's like, yeah, yeah.

[Interviewer]

Where you and I branch is that you keep thinking your base assumption clearly stated that humans.

[Andy Weir]

Yep.

[Interviewer]

Humans are always going to maintain control. My only thing is, and again, for me, this is just a question mark. Will humans be able to control it? I don't know. We both end up in an optimistic stance. I'm certainly not trying to convince you, and you have been exceedingly clear.

[Andy Weir]

Well, you talk about, will humans be able to control it? Like, I mean, unless. So ultimately it comes down to humans. Again, if a group of humans has enough power like a government to give AI executive authority, that might turn out to be a bad idea. But it's humans doing that to humans. It's humans saying, I'm going to give AI this power, these powers, and then maybe it backfires. But it's ultimately it's humans who make that decision. The only way I, I can see for your scenario to play out is, I think, correct me if I'm wrong, you're talking about an AI that's so smart, it can manipulate people into doing what it wants without ever having executive action. It says, like, it can convince this guy that he should do this thing. It can convince that guy that he should do that thing. It can, it can talk you into this and you into that. And you know, now you, now it's basically using humans as its executive action by being able to convince them of things. Okay, so that's a, that's an interesting concept. I like it. But then you're just talking. In my view, what you have here is an AI politician. You have an AI that's trying to convince large groups of people of an ideology that it has devised that it thinks would be good. And if it can do that, why would it need destructive. Why would it need destructive executive authority? If you make an AI and you say, like, I want to make the world better, and it's like, okay, I convince everybody to share, you know, that's the quickest way to do it. I don't know.

[Tom Bilyeu]

Taking a quick break. But there's more with Andy Weir when we come back.

[Commercial Announcer]

When you manage procurement for multiple facilities, every order matters. But when it's for a hospital system, they matter even more. Grainger gets it and knows there's no time for managing multiple suppliers and no room for shipping delays. That's why Grainger offers millions of products in fast, dependable delivery so you can keep your facility stocked, safe, and running smoothly. Call 1-800-GRAINGER click granger.com or just stop by Granger for the ones who get it done.

[Interviewer]

All right, we're back. Let's get into it. All right. Going back to space travel, which is something that you're super passionate about given AI, given the advancements that we are likely to see there, let's say, in the next 25 years, how, as the person who wrote the Martian, how optimistic are you that we're actually going to get to Mars, start colonizing it? Do you really see that as a thing that we'll see in our lifetimes?

[Andy Weir]

I think it's only a matter of time before we go to Mars. I al. I also think people underestimate this one aspect of, like, space travel, which is that we think of space travel as, as, like this endeavor that we have to do, but it's proven to be economically unnecessary, right? It, it, it's hard to make an economic case for why we should go into space. You can make a survival of the human race and case you can do that. The only economic approach I've ever come up with, Earth base, is simple tourism, right? So there's nothing, there's nothing you can mine on the moon or Mars that you can't find easier on Earth or. And if you did want to mine it or collect those resources, we are now in an era where you could do it all remotely. Like you don't need to figure out a way to keep squishy humans alive and make sure that they can get back. Right? So I. But if the technology comes about to make it so that it just doesn't cost that much to go, then people will do it. Just because humans have a Wanderlust. So one analogy I like to use is saddles. Like a modern saddle that you put on a horse is the most, I mean, people in the 1800s would kill from modern saddle. It's made out of synthetic materials that don't rot or don't have any problem with getting wet. It's made to be comfortable for the animal, made to be comfortable for the person. It'll last basically forever. It's incredibly safe. It, you know, it just has all these features that they just didn't have in the 1800s. Now in the 1800s everybody needed a saddle and everybody used it. Nowadays it's, you know, hobbyists, enthusiasts, athletes, you know, stuff like that. But it's not a thing that's a core aspect of human need anymore. It's not a basic thing that we all need. Like if no one was able to ride horses anymore for the rest of time, it would not have a huge effect on human society anymore. Right? It would have if that happened in the 1800s, but now it doesn't matter yet. If you look at a modern saddle, it is this incredible feat of engineering. But you got to bear in mind every aspect of that was not invented by, by a national push to make a better saddle. Right? Like we have the, you know, National Aeronautics and Saddle Administration, right? We didn't go out of our way to invent carbon fibers to make lighter saddles. We didn't invent nylon to make better saddle straps. We didn't invent computers so that we could model straddles saddles so we could, you know, make them better. We didn't invent 3D scanning so that we could scan horses to make better saddles. We didn't invent any of those things for the saddle industry. But they made use of the technology that was developed for other reasons to make better saddles. So I think that's what's going to happen in the space industry. I think what's going to happen is yes, there are companies like SpaceX and Blue Origin and a bunch of others that are going out of their way to try to do space travel stuff. But I think even if those companies didn't exist, the technology would come about through the normal process of, you know, profit driven technological development that ends up being really good for space travel. Like somebody be like, you know, we end up, you know, jet engines end up breaking down a lot because the heat, heat, heat warping of the turbine fans is what causes metal fatigue. And if we could find a way to make that heat warping not happen, we Might be able to save a lot of money on maintenance. And, and so they do a bunch of research and they're like, okay, we figured out a material that won't heat warp anymore and we're going to make our jet engines out of that. Then meanwhile, somebody at the rocket industry is like, I could make a rocket engine out of that that won't melt. You know, that's, that's the kind of way I see that the, that space advancement in space technology will go. It's like secondary accidental benefits to the industry

[Interviewer]

does what's going on.

[Andy Weir]

And then once it becomes cheap enough, once it becomes cheap enough because of all this other, because of all these technologies that humanity invents for other reasons, making it so you can put together a rocket, then, then once, once it's cheap enough for like ordinary people to go to space, people will just do it. Because humanity has, I believe, a genetic desire to spread out. I think that's actually an evolved trait. So that like one, you know, one flood in one valley doesn't wipe out our species. That makes it, we, we have a desire to fan out into other biomes and go everywhere. And that makes it pretty much impossible for us to be rendered extinct even by things like a, you know, the near extinction event about 75,000 years ago killed a lot of stuff, didn't kill us because we were all over the place.

[Interviewer]

Yeah, that, that was going to be the button that I pushed on is given what's going on with Blue Origin, with SpaceX, you have a generation of kids that grew up wildly inspired by the space program landing on the moon. They were, you know, sort of bothered by the fact that the space program stalled out and decided that they were going to do something about it. Do you think that, I mean, it seems like it's already driven some pretty radical advancements given that Elon Musk and SpaceX now account for 95% of the tonnage globally that goes up into space. Given that Elon has said that he is going to build a colony on Mars, I think Bezos has his eyes all over the moon. Whether they have a purely economic reason to do it or not, given that they have so much money and in the case of SpaceX have built a truly thriving self sustaining economic engine, do you not think that will lead to these breakthroughs coming at a faster pace?

[Andy Weir]

It could certainly help a lot. Like there again, I always go back to the aeronautical industry. It's like things go really fast at first once. So the true invention of SpaceX wasn't so much rocketry, it was how to make money on rocketry. Because all the other thing, you know, the other companies that were contractors to make, you know, spacecraft for the US Government and for European Space Agency and so on, they were not interested in driving costs down or competing with other companies. They were just interested in getting the job done. So what SpaceX's real innovation was they said they had an obsessive focus on driving down the price so that they could take over the market. Because no, before SpaceX nobody looked at spacecraft as a market. But what SpaceX saw is that there is a total addressable market for spacecraft. It, the customers are almost 100% governments and with the occasional very large company that needs satellites like communication companies. So it's a very small number of potential customers. But they saw that this is a possible for profit business if you do it right and you take over the market and almost all of the total addressable market goes into your hands. And I think that's what SpaceX did. But something like that can't survive. A business I don't think can survive if it relies entirely on discretionary government funding. And which is what the vast majority of SpaceX's money comes from is like space entities and like NASA, ESA, JAXA, that sort of thing. I, I just, if you don't provide a service to consumers, I don't think your business is going to survive for very long. So that's why, and I think Musk knows that and I think that's why he's trying really hard to find a way to drive the price down to the point that you can have direct consumers like, so that you, Tom, could go into space, you've got a lot of money, but someone else who doesn't have that much money could go into space, you know.

[Interviewer]

Now what about like what he's doing with mining of asteroids? Do you think there's anything there? Is that pipe dreamy?

[Andy Weir]

I don't, I mean it's not pipe dreamy. I think the only real benefit to mining asteroids is you're not doing ecological damage on Earth. So. But being environmentally conscious rarely pays off financially. So I think unless again you have government largess at play, I don't think that's worthwhile. So if a government wants to say, hey, we want a bunch of, you know, lithium, but we don't want to strip mine, you know, some location on Earth because that's unpopular. So people elected this government to be environmentally conscious. So we are going to help fund methods of acquiring lithium in a way that doesn't disturb Earth's environment, then there's a potential market for that, but not a very big one. Because economics always finds the easiest, cheapest way to do things. And then morality finds a way to justify it after the fact. And so eventually it's like, I mean, some of the most human rights activist people I know are the people who own iPhones. Right. I mean, like your phone was made with slave labor. Why aren't you worried about them? Well, that's different. Like, yeah, because morality always finds a way to justify the economics. It doesn't work in the other direction very often.

[Interviewer]

Yeah, that is distressingly insightful. What is? What is?

[Andy Weir]

Well, I mean, if you want to get even darker, I mean, humanity only suddenly realized slavery was bad after the industrial revolution when they no longer really needed slaves to do mass production anymore.

[Interviewer]

Oof.

[Andy Weir]

How's that for an observation?

[Interviewer]

That's rough.

[Andy Weir]

Like, that's rough. 10,000 years of slavery in every human culture everywhere. And it was only about mid-1800s when suddenly humanity's like, oh no, this is wrong. Well, what else happened right around that time? The ability to do these things without slaves.

[Interviewer]

Wow.

[Andy Weir]

That cheaper and more efficiently so that it was no longer an economic need. Yeah. So I have my dark. I'm not just a pure Pollyanna. I have my dark observations too.

[Interviewer]

That's amazing. Well, I think a big part of what everybody loves is the optimistic side that comes through in your writing very, very aggressively. Swinging us back to the optimistic side. What is happening right now that excites you the most in terms of like, technology? Whether it's AI, whether it's space travel, like what's got that thing that really has your imagination spinning.

[Andy Weir]

Boring though it is. Self driving cars. That's the thing that I'm really excited about. I think people are all dramatically underestimating the tremendous impact it's going to have on our society once we get self driving cars perfected to the point that they are legal everywhere, affordable by a typical consumer, and just found everywhere. I just don't think people fully understand how much it's going to affect our society. So consider some things. First off, some downsides. The entire driving industry is going to go away. There won't be any more truckers, cabbies, none of that stuff. Or truckers will really be more like directors. They'll be like a person who sits in the truck and then makes sure that the cargo is received on the other end or something like that. Or there might even what used to be a trucker might be a guy who's stationed somewhere to make sure that the truck deliver whatever. So you're not going to have any of that anymore. The next thing is you're not going to have to park anything anymore. Cars will just drop you off and then go away. You don't have to park a car anymore. Something like what? 20% of the surface area of any given city is parking. That's all just going to not be necessary anymore. Think of, I don't know where you live, but are you in a city? What? Where do you live? LA? Yeah, LA. Think of how many parking structures, parking lots, parking garages, street side parking, all that stuff gone and, and used for other things. Okay, next up, 50 to 60,000 people a year won't die in drunk driving accidents because you can be as drunk as you want when you go home. Next up, cities are, the transition is going to be rough because everybody that relies on driving like cabs, trucks, on stuff like that are going to be against allowing self driving cars because they're going to be like, this is eradicating my profession. I don't like it. Cities, very powerful civic entities like Los Angeles for instance, are going to be against it because LA derives over $100 million in revenue per year out of traffic. Citation. Get rid of those. It just loses traffic and parking. Citation, LA did. I remember it was about 10 years ago, LA was their fire department wanted GPS systems in their trucks so they could more quickly get to, you know, emergencies. The city said, we don't have the budget for that. And then in that same year, the city replaced all the parking meters with automated parking meters that can alert parking enforcement if you say even one minute over the time. And that cost them a bunch of money. But see, that's a revenue stream. Every entity will always orient itself around the revenue stream, no matter what it's supposed to be originally. So cities are not going to like the idea of just losing all that revenue and they still have to maintain the roads, right? So it's not like they're losing a cost or an expense, they're just losing the revenue. But getting past that because like I said, economics always wins. We've seen the taxi cab industry already get destroyed by Uber and Lyft right now. Moving on, why would you own a car? Why wouldn't you just subscribe to a car? Service is just like Uber, but driverless cars, if it's quick to. I mean, eventually owning a car will be kind of like owning a horse. It'll be like, you don't, you don't need that. And I mean, you might want your own private car because you don't want the stink of someone else has been in your back seat. Okay, It'd still be driverless, right? You just get in. And then for people who don't own cars, they're like, people in 100 years are going to think it's absolutely absurd that we dedicated about a third of the interior space of our house to a place that stores a car. Like garages or like a big chunk. If you're, if you're just living in a suburban neighborhood, your garage is like a third or a quarter the total surface area of your house. Like now people will be like, you're going to see a lot of family rooms all of a sudden. You know, why, why, why own a car? Why maintain a car? Why get a driver's license? Why would there be a driver's license? Do we need a dmv? Yeah, somebody's got to register the electric cars. But that's a business thing. That's like a, that's kind of like registering your freight train, you know, it's like a thing where the business talks to, talks to the government to get the permits necessary. It's not a thing that affects the layman. No more cars, no more driver's licenses, no more automotive insurance. A lot of people in the insurance industry, which is very powerful in America, are not going to like the idea of people no longer needing insurance. Because of course, the responsibility for car crashes won't be on the person who was in the car. It'll be on the company that made the car.

[Interviewer]

No, it's going to be wild like that.

[Andy Weir]

Massive, massively disruptive in a good way, for the most part. Also agreed. Also for the environment. It'll be good too, because people, there'll just be fleets of electric cars. You won't need to be able to go hundreds of mile. It'll be like, okay, there's an electric car will come pick you up, drop you off, and then I'll go pick someone else up and it'll go recharge when it has to. You don't need gasoline anymore. Okay. If you're going to do a long haul trip, maybe you do. I don't know.

[Interviewer]

Yeah, that'll be interesting, especially to see if truly electric cars can get off the grid right now, which they're still tied to that. But whatever, we'll assume that problem will get solved. I want to go to. So I know your latest book. It's my understanding from things you said publicly that your New book integrates AI in some substantive way. Without talking specifically about your book, what made you want to tackle the issue of AI?

[Andy Weir]

So, full disclosure, that was the book I was working on, but I shelved it, I backburnered it, and I'm working on a completely different project now. What made me want to do it was it's just the thing that I thought was a cool concept. This was before ChatGPT came out. This was before all the AI stuff exploded. And so I was just like goofing off. And it was an idea I had and it was novel and unique. Honestly, part of the reason I've chucked or I've back burnered this book is because now it's in the public consciousness so heavily, it would seem like hopping on a trend. So interesting. There are going to be a million science fiction books about AI over the next few years and I don't want to just be one of those. So I'm like, well, when I was working on it, it was a unique concept, but by the time this comes out, it won't be. So I'm going to do something else. It's just going to become too saturated. So I was a little bummed. But yeah. So I actually decided to get rid of it. But it was my usual approach of like, okay, I'm going to look at this realistically, not like, yeah, there's no rogue AI that tries to take over the world or kill everybody. Right. It was just like how people are using AIs that matter now.

[Interviewer]

Are you talking themes or anything about the new book?

[Andy Weir]

No, the new book. I'm keeping that close to the vest now too. Sorry, I'm not going to talk about that yet.

[Interviewer]

As a huge fan, whatever you're working on, I absolutely cannot wait to see what happens. Where can people follow along with you until that next one comes out?

[Andy Weir]

I mean, I have social media, Facebook is Andy Weir and Twitter is hashtag or sorry, X is Andy Weir, author. Sorry, I didn't mean to dead name you Twitter. Yeah, but I'm not like crazy super active on those. It's just when I have an observation like, dang it, why have we had two bottle episodes of severance in a row? Then I'll post that sort of content.

[Interviewer]

I love it. Well, Andy, I can't thank you enough for joining me today and everybody at home, if you have not already, be sure to subscribe. And until next time, my friends, be legendary. Take care.

[Andy Weir]

Peace.