Things Aza Raskin Has Said on Podcasts

“In some sense, the fundamental question we need to answer as a species about whether this episode is the season finale or the series finale of our species is whether we can escape game theory. Do we do the thing that game theory says? And what is game theory? It's sort of like the ability to know what is the strategically right move.” — Aza Raskin, StarTalk Radio

Chuck Nice has just asked how Aza squares building guardrails on AI with the risk that other countries won't, leaving a rival ahead economically and militarily. Aza answers that the core question is whether people can escape the strategic logic of a race.

StarTalk Radio · 2026-10-30 Permalink → Listen →

StarTalk Radio Around 37:06 into the episode

Speaker 1

Aza, on your resume is the dubious invention of the infinite scroll, where we all got caught up

Chuck Nice

…on, but does that mean other countries will as well? And if they don't, that puts us at an economic and ultimately perhaps a military disadvantage to them if they let AI completely run ahead of us. So where does your thinking and your actions, how do they land in the middle of this conundrum?

Speaker 2

Yeah. In some sense, the fundamental question we need to answer as a species about whether we will, whether this episode is the season finale or the series finale of our species is whether we can escape game theory. Do we do the thing that game theory says? And what is game theory? It's sort of like the ability to know what is the strategically right move. In the face of

Chuck Nice

someone else who's also playing that same game.

Speaker 2

Exactly. That's exactly right. So you're saying it exactly right. Like if I don't develop like the most powerful AI and my adversary does and they can cyber hack me or if I'm hitting fast forward more on technological progress, I'm making new weapons or my GDP is out competing yours, then you will…

Chuck Nice

theory also includes the possibility that it may be that if we cooperate with our fellow players, everyone is in a better place. That's right. That's a possible solution in some configurations of game theory.

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.

“We sort of think of social media as a kind of baby AI. All it could do is rearrange human posts and human content, decide what we see, and a tiny little misalignment, where it wasn't optimizing for our well-being or our health or for truth or for knowledge, but optimizing for getting our attention, that was enough to have destroyed much of the world we love the most and the highest rates of teen suicide, things like that. And the question to ask is, Charlie Munger famously said, show me the incentives and I'll show you the outcome. Have we changed the incentive as we move from baby AI to mammoth AI? And the answer is, of course, no, we haven't. So why would we expect any different outcome?” — Aza Raskin, StarTalk Radio

The discussion is about the attention economy and how engagement-driven feeds were built. Aza describes social media as an early, limited form of AI and argues that the same incentives now apply to far more powerful systems. The causal claims about social media's effects are his opinion, not something the transcript establishes.

StarTalk Radio · 2026-10-30 Permalink → Listen →

StarTalk Radio Around 08:15 into the episode

Speaker 2

…the race to the bottom of the brainstem where you have supercomputers that are at that point, like the largest supercomputers in the world were pointed and trained at your brain trying to make sure it knew every part of your social psychology and vulnerability to keep you engaged and attended to.

Chuck Nice

Wow. Okay. So a lot of good that did.

Speaker 2

Well, you know, we sort of think of like social media as a kind of like baby AI. Like all it could do is like rearrange like human posts and human content decide what we see and a tiny little misalignment where it wasn't optimizing for our well-being or our health or for truth or for knowledge, but optimizing for getting our attention. That was enough to have destroyed much of like the world we love the most and the highest rates of like teen suicide, things like that. And the question to ask is, you know, Charlie Munger famously said, show me the incentives and I'll show you the outcome. Absolutely. Have we changed the incentive as we move from baby AI to like mammoth AI? And the answer is, of course, no, we haven't. So why would we expect any different outcome?

Gary O'Reilly

Is this social media, the sort of birth of social media, the birth of the attention economy? And this has been the total focus of the creators of social media and AI?

Speaker 2

Yeah, I mean, obviously there's always been advertising and things like that, but there is nothing that sat in your pocket. So this sort of extends the work of like Neil Postman, Marshall McLuhan. But when an AI knows you, and this is why whenever we're talking with parents, we say, you know, you…

Chuck Nice

Like you said, it really isn't the beginning. It's kind of, I'll say, the perfection of the making of the attention or the manipulation of the attention economy. Because ever since right after World War I, when propaganda was discovered to have very beneficial effects in bringing people as a…

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.

“There should be a penalty. The longer that, say, a company like OpenAI doesn't disclose or doesn't find something that their rogue AIs do, the higher their penalty should be. And it shouldn't just be in money. It's sort of like, if you're a soccer team and you're doing bad behavior, you get a red card. You have to go down a player or you have to go down two players. Well, what's the penalty in AI? Well, it's like take away some of the compute. You get a red card… You just can't go as fast anymore because you're not doing it safely.” — Aza Raskin, StarTalk Radio

Aza is listing policy changes he thinks should happen right away, after proposing that frontier-technology companies publish the ways their products could go wrong and have those lists peer-reviewed. Here he proposes escalating penalties, paid in computing power rather than only money, for companies that fail to disclose problems with their AI systems. This is his proposal.

StarTalk Radio · 2026-10-30 Permalink → Listen →

StarTalk Radio Around 1:08:19 into the episode

Speaker 2

…things can go wrong. And that should be peer-reviewed. Everyone else, and actually there should be sort of a bounty program. If I can find a harm and I can think through to a harm that you didn't name, you should have a penalty. So we want to like maximally incent everyone to really think through.

Chuck Nice

So the unknown unknowns would bring about the penalty.

Speaker 2

…doesn't find something that their rogue AIs do, the higher their penalty should be. And it shouldn't just be in money. It's sort of like, you know, if you're a soccer team and you're doing bad behavior, you get a red card. You have to go down a player or you have to go down two players. Well, what's the penalty in AI? Well, it's like take away some of the compute. You get a red card. I'm sorry. You just can't go as fast anymore because you're not doing it safely. And so I think there's some really, really sensible things that we can do. Like right now, these agents hack out, but who's responsible? And there's this weird sort of game of, well, it's the AIs did it. So that means a human isn't responsible. But no, a human being hit the button that said, let's start that. Like there should be forms of criminal liability for executives for when the agents…

Gary O'Reilly

Don't the European Union impose a fine of up to 3% of their global turnover as a penalty? I mean, is that the sort of parameter you're looking at?

Speaker 2

Yes, but I think the punishment or the fine should come in the currency that matters most. So,

Gary O'Reilly

right,

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.

“Imagine a chimpanzee being like, okay, so there's going to be a new version of me that's like a bit smarter. What is the risk? Well, is it going to like eat more of the bananas? Like, well, what are they really going to do? Chimps could not conceptualize the atomic bomb and like roads. And this is the same thing. We cannot conceptualize the actual risks of smarter than human intelligence… because it is just beyond our comprehension.” — Aza Raskin, StarTalk Radio

The conversation is about whether a far more intelligent AI would compete with humans for resources, and Chuck Nice has joked that chimpanzees would be our enemies if they knew we liked bananas. Aza takes the analogy further: a chimp could not imagine what a smarter version of itself would build, and humans are in the same position with smarter-than-human AI.

StarTalk Radio · 2026-10-30 Permalink → Listen →

StarTalk Radio Around 1:05:05 into the episode

Speaker 2

…question you need to ask is earthworm or chimp is: are those other more intelligent beings, is AI, is it competing for the same resources as we are? And if the answer is yes, you know, it's going to go very poorly for us. If the answer is no, then we just get stepped over and not stepped off. Yeah,

Chuck Nice

if chimpanzees knew that we liked bananas, we would probably be their enemies. Yeah,

Speaker 2

…so there's going to be a new version of me that's like a bit smarter. What is the risk? Well, is it going to like, are they going to like eat more of the bananas? Like, well, what are they really going to do? Like, chimps could not conceptualize the atomic bomb and like roads. And this is the same thing. We cannot conceptualize the actual risks of smarter than human intelligence in these swarms because it is just beyond our comprehension. But I do want to go back just for one second to something I find really beautiful here. Because the fundamental question is: okay, so you cast the light of the human mind through the world and the shadow is language. And we're going to do the same thing with every communication system for whales and for parrots and for elephants. And the question is, how much do those shadows overlap? And the…

Chuck Nice

mathematically, we would describe these as other dimensions of information. Yes, yes, yes, yes. Right. There's the shape and then there's some other aspect of that shape once you have the shape. And so, okay, proceed. No, it's exactly right.

Speaker 2

And so you then take the shape for English and you take the shape for images. And it's not quite a rotation anymore, but there's a pretty clean mapping that just puts these two things together. And now you can translate from language into images. But it's not just that. It's also code has a shape.…

Chuck Nice

just to be clear, you could not have done that if you didn't otherwise know what they were looking at. You would just have this fMRI signal. The connectivity is because you gave them something to look at. That's right. So you've established the correlation

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.

Things Aza Says on Podcasts

Where to Find Them

What They Said

Collections They Appear In