The Dispatch Podcast - Will Artificial Intelligence ‘Kill Us All?’

Episode Date: September 15, 2026

Steve Hayes is joined by Jonah Goldberg, Mike Warren, and Klon Kitchen to discuss the resignation of an Anthropic researcher over concerns that AI may “kill us all,” yet another flip-flop by the Trump administration on Ukraine, and President Trump’s $5,000 bribe to voters during the midterm convention. The Agenda: —Jacob Coxon's resignation tweet —Pacing the AI frontier —The Hugging Face incident —AI and foreign policy —Russia strike near David Petraeus —NYWT: Trump bribing voters Show notes: —James Miller: Why AI might kill everyone Learn more about your ad choices. Visit megaphone.fm/adchoices

Don't you wish you could just hit skip on the worst parts of your life? You know, the same way you can skip an ad? I get it. I'm Siaia and I live in Ice Cove. I've made some questionable decisions that didn't end up the way I planned. And today, I'm still figuring it out. Somehow things usually get worse before they get better. Apparently, that's how I roll.

So bundle up and come along for the bumpy ride. Stream a new episode of North of North Tuesdays on CBC Gem. Welcome to the Dispatch podcast. I'm your host, Steve Hayes. On today's podcast, we discussed the end of humanity. No, really. Last week, Anthropic researcher Jacob Coxon quit the leading AI firm, arguing that, quote, the people-building AI earnestly believe that it could kill us all by the end of the decade, unquote.

His comments, supported by other leading AI researchers, kicked off a feverish debate about the future of AI and humans. We take a careful and sober look at that debate and the issues involved. Then, the Trump administration over the weekend blamed Ukraine for prolonging the war Russia started nearly five years ago and for creating a diesel fuel crisis with its attempts to defend itself. What to make of those claims? And finally, not worth your time. The president stakes the midterms on a $5,000 vibe. I'm joined by Jonah Goldberg, Dispatch co-founder, Mike Warren, politics editor at the dispatch,

and Clon Kitchen, who for 15 years served as an intelligence. officer focusing on counterterrorism, cyber operations, and covert actions. Klon was national security advisor to former Senator Ben Sasse and played a pivotal role in the U.S. Cyberspace Solarium Commission, crafting the strategic blueprint for America's cyber defense. Most significantly, Klon is the former author of the Dispatches Tech and National Security Newsletter, The Current. Let's dive right in. Welcome, gentlemen.

The latest debate on AI started with a tweet from Jacob Coxon, an AI, an AI- researcher from Anthropic who had spent three years at Open AI before that. Coxon tweeted last week that he was quitting the company, Anthropic, because of his concerns that the race to win on, quote, self-improving superintelligence has the company's acting irresponsibly. As of Monday morning, his lead tweet in the thread has 171 million views. A couple of brief excerpts from his thread here. I resigned from Anthropic today. I spent the last three years doing pre-training research at both Ophabwe.

Open AI Anthropic, neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. Do not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources. We've all witnessed the progress in each of these domains and progress is not slowing. The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt. If anything, many executives and senior researchers will couch their phrasing in the press

to sound sensible. But I hear the same people express fear privately. No other human activity poses this level of danger. Klon, immediately after he tweeted that, several other senior people in AI safety research echoed his claims, including Evan Huberinger, a team lead at Anthropic Alex Turner, a former research scientist at Google DeepMind, and Jakub Pachshon-Chalksky, chief scientist at OpenAI, urging extreme caution. Klon, you spend a lot of your time these days looking at AI, thinking about AI, talking about

AI with the people driving these developments. When you hear this, the people building AI earnestly believe that it could kill all of us by the end of the decade. What do you think? Well, I think Jacob is accurately representing a significant portion of the AI safety group out there. And look, there's enough capability that's been on display, which I can talk about in detail, and enough concerning developments to continue to justify some of these concerns. But not everybody feels equally alarmed by what Jacob is highlighting.

Let me just begin with kind of some of the presuppositions behind what he laid out. The particular advancement that is driving what he wrote about is what's called recursive self-improvement, RSI. It's the idea of AIs building themselves, right? And so the fundamental concern is up until this point, as we have been building artificial intelligence, the dynamics that we've experienced have been increasingly where we understand. understand the inputs, we understand the outputs, but what happens in that magic middle is kind of a black box. Like, it's just looking at a volume of information and making connections and then doing interesting things with that information. That is very difficult for us to understand and articulate. That's where we are now. At the point where AIs are able to successfully build and

refine themselves, we start to lose visibility on not just what's happening in the middle, but perhaps even the inputs. Because we're not doing it, because this is not us anymore. That's right. It's doing it on its own. Precisely. And so the concern, that's what's called takeoff. So at the point where these AIs are building themselves and we, the humans, lose visibility

on kind of that whole process and all we're really dealing with is the outcomes, that's what gives people concerns. And so they take that kind of technological trends. which we are definitely making very real progress toward recursive self-improvement RSI. And that is now in view of being a possibility, frankly, the timeline keeps shrinking. But that is then coupled with the sustained exponential growth and broader capability of these systems in terms of everything from cyber capabilities, mathematics, biotechnology, and biological research, and a whole lot of other things.

And so those two things get wedded together. in the minds of AI safety personnel, and they say, okay, so these things are getting better and smarter and more capable at a rate faster than we even anticipated, and we've been very aggressive in that estimate. And we're coming close to a point where we may lose visibility in terms of how these systems are being built and working. And those two things combine to create a moment of concern, right? So I think what I've described there is a cost. commonly accepted state of the practice. I think, you know, then this conversation begins to proceed from kind of first principles.

You know, it's really worldview driven, right? So if you think that these systems can become sentient and synthetic neurons are no different than biological neurons, and so there's no reason to think that the evolutionary process wouldn't then produce these kind of living machines, well, then you quickly see how that kind of leads to one endpoint. But then there are others who say, okay, I don't have to kind of swallow that whole worldview. It is sufficient to say that we are building machines and systems that are very capable and that we have innovated faster than we can secure.

And this is going to have massive social implications. And, you know, Jacob and Dario, I think, are... This is Dario Amadee, who's the CEO of Anthropic. they are accurate avatars for a part of that spectrum of worldview. And then someone like, you know, maybe Sam Altman and others would be lesser variations of that. And then you've got Elon Musk that depending upon the ex post, you know, kind of straddles the line. So I could keep going. But like that's a scene setter for what Jacob was talking about and where he's coming from.

So let me go a couple clicks deeper on that. You've characterized the views that Jacob has expressed as characteristic of at least part. of the AI world, the AI safety world. I think the thing that made so many headlines, in addition to the fact we should point out, Jacob Coxon then went on and gave interviews to virtually every major media owl and known to man and was sounding the alarm.

He seems genuinely alarmed by this. Certainly knows a lot about this. But I think the thing that captured the attention of people, and this is no doubt, I think, why he framed it the way that he did, was this end of humanity in a decade. And we've heard these things. If you followed the AI debate before, the DOOMers have said these kinds of things before.

But I think it was the fact that Coxon tweeted this out that he had immediate affirmation in rapid succession. I mean, this all happens within 15 minutes. There's a Wall Street Journal story. You have some people who are skeptical of the AI safety types calling this an op, saying this was rolled out. This was meant to alarm. It's designed to get the kind of attention that it has. they deny this. But what of that claim that this could end humanity? How would that even happen? And if it's possible, are you going out and now eating, you know, donuts for breakfast

and getting cheese on your burgers? Well, yes, to the donuts for breakfast and cheese on my burgers, but I've been doing that for a long time. It has nothing to do with AI. So if I were to kind of steal man the argument, why someone said, I don't think you have to go full Terminator, you know, know, Judgment Day, Rise of the Robot. Steve, that's a movie with Arnold Schwarzenegger where an artificially intelligent computer called Skynet tries to destroy humanity. I thought you should just know that.

I'm sorry, go ahead and call. I have never seen Terminator, actually, so I appreciate the plot summary. I wasn't being a jerk. I was... I mean, you were both. You were being... Fair, fair.

If a running back's name comes up, you'll do the same thing to me, so... Fair. So look, what Jacob said was, there are people in these labs who think that we could destroy humanity in the next decade. When he says that, he is absolutely correct that there are people in these labs, very influential engineers and other leaders, who do believe that. I mean, this is not a minority view. I don't know that it's a majority view, but it's not a minority view within the AI research space. I mean, this is not, they are not representing a crazy fringe.

who's just alarmist for the sake of being alarmist. There's a serious group of people who hold a view like this. Can I contextualize that question just a little bit differently? When you say there are serious people and engineers in there who believe these things, so it's not the cafeteria lady, right? It's real people. Okay, fine, grant you that. I can say there are people in the Trump administration

who believe that Vladimir Putin is a foremost champion of Christianity and traditional values, right? that would be accurate. It would also just be really wrong and more of a comment on the ideological valence of some whack jobs in the Trump administration. Similarly, there are a lot of people in Silicon Valley who are transhumanists. You know, there is this weird, culture-ish thing that you should always be a little careful when someone says humanity is doomed and I'm the prophet telling you how to avoid it, right? So there is a kind of religious thing going on in all sorts of ways.

And so I don't want to speak in broad stereotypes or anything, but in the world of incredibly high EQ, socially aware, nuanced men of the people, the bleeding edge of the top point O.O.O.1% of computer programmers are unrepresented in those groups, right? They are not people persons. And so I'm just wondering how much of it is, it's not that they're freaking out because they're nervous Nellies,

it's that they are conjuring these scenarios because they have just a really atypical bunch of views about how we're going to hit the singularity, or but I have this Promethean power to create our own replacements. I mean, there are all of these things that go on in that world too, and I'm just wondering how much of that is skewing things. Yeah, and this goes to my point about, like, At some point, this just comes down to your priors.

I would say that when Jacob points to this, there is a fringe out there, right? There's just different flavors of this, right? So there are people who are worried about the singularity and who are transhumanists. And I would put that further to the fringe. But let me give you an example of a more serious argument that I still would call low probability but high impact scenario. So when we look at some of the places where AI has had the most significant impacts, it's been on things like biological research. So protein unfolding and things like that from DeepMind.

DeepMind is an AI lab, part of Google. They have an AI model called AlphaFold, and it has essentially done two centuries worth of biological research in six months when it comes to protein unfolding. And if you can know how to kind of fold and unfold proteins, you essentially know how to structure biological life. So you take advances like that and you start building scenarios and you say like, okay, well, if the models themselves have this capability and if the trend thus far has been that these capabilities are now proliferating rapidly so that anybody can do it, Well, then the risk of a significant biological event, particularly if technologies like CRISPR and other gene modification things

also become more democratized. Well, then the risk of, you know, a catastrophic biological novel pathogen being created and released into the world, that has materially increased. And there's a book, I've forgotten her name, there's a book that's just out. It's called Biow War. I'll think of in a second.

Maybe we can put it in the show notes. But this journalist does this interesting report where she explains what that could look like in the case of a novel pathogen being released into the United States. So there are some people who look at things like that. And they say, look, there are governments out there. So I'm not necessarily worried about the guy in a garage,

but there are governments out there. I mean, we could think of one, I'm sure, who would certainly seek to use these technologies to build these kinds of novel pathogens for their own purposes and who don't have a spotless record in terms of securing those pathogens. And, you know, hey, look, we're actually rocketing toward a real catastrophic event. So there's a flavor of people, and I think that's probably the majority without polling.

I think that's largely representative of the majority of concerns. But there are others who take it much further, as Jonah was describing. So I'm not trying to do either or here so much as I'm just trying to explain that this capability has advanced so quickly and so expansively that we've actually, unlike most technologies, we've actually been underestimating how capable these things were going to happen or become and how quickly they were going to become that capable. So that it's causing even people like me, who I would say generally I'm optimistic on the technology, to say, okay, we have a new kind of challenge here. And what I'm discovering is, is that government and governance more broadly is simply not keeping up.

All right, we're going to take a quick break, but we'll be back soon with more from the Dispatch podcast. This family is on the brink of civil war. On September 18th, Mobland. The hit original series is back on Paramount Plus. We are the Harrogans. Don't know the yet? And Google us. From the underworld of Guy Ritchie.

Do you want to step up the ladder? I won't call him dead. Starring Tom Hardy, Pierce Brosnan, and Helen Mirren. Do I have to do everything myself? You want to want. I'll give you a... Mobland. New season hits September 18th on Paramount Plus.

And we're back. You're listening to the Dispatch podcast. Let's jump in. So I want to get to government, governance. But two points first. One, last week, Dario Amadeh, the CEO of Anthropic, released a long statement about the kinds of things that these leading AI companies should be doing. Some of them on their own, some of them in collaboration with,

other companies, some of them in collaboration with governments. It was a call in a sense to be regulated in addition to announcing what Anthropic itself would be doing on its own to protect against some of these concerns. He also, Anthropic also released basically an internal intelligence report detailing some of the things, Klon, that you just mentioned there, which is to say, hey, we have seen Claude being used to try to advance bio-weapons research by some you know, countries that we would probably not prefer to have doing this kind of research. Klon, let me come back to you one more time before we bring in Jonah and Mike about a broader discussion with a specific example that I read over the weekend and get you to respond to that.

This comes from James Miller, who's a professor of economics at Smith College. He is a PhD from University of Chicago as his law degree from Stanford. Three paragraph social media post. He writes, a short explanation of why AI might kill everyone. AI may soon be much smarter than humans. When it is, it will be in control. If you were put in a classroom of kindergartners, you would be the one in charge. Once AI has the robots to build and maintain everything it needs, we will become useless to it. But keeping us alive will oppose enormous costs. We need clean air, water, and food, the fastest, cheapest ways to build more computing power could produce enough pollution to poison the biosphere. Humans have a reason to avoid that. We depend on the

biosphere. AI won't. Unless it values our survival, preserving the conditions we need to live will be a cost it has no reason to pay. Building what it wants as cheaply as possible would mean poisoning the biosphere and killing everyone. He concludes, we're not programming AI's values line by line. We're training and without really understanding what it is learning to want. We're giving control to something we don't understand, something that will have no intrinsic need for us, and something that will easily be able to wipe us out. This is crazy, but it's what we're doing, even as some of the people building AI tell us it's crazy. Klon, how do you respond to that? Is that hysterical? Is that alarmism? I think his first two sentences do a lot of work in terms of it's in control. Well, hold on.

You just skipped a lot of things there, right? And I think it quickly begins to anthropomorphize the technology that I think illustrate some of his priors, right? So I would at least question if not reject some of his kind of assumptions as he kind of draws out that scenario. So I think he and I would have differences in terms of what I think the threat is. But that doesn't mean that I think there's no threat. I just think there are more serious, proximate concerns that, frankly, those more hyperbolic scenarios distract us from, right? So, for example, you mentioned the Anthropic Report. Well, the gist of the Anthropic Report, it said a lot. It's 300-something pages. But essentially,

what it said was, the cyber capability that was formerly reserved only for the most sophisticated nation-states has now been fully democratized, and we have seen it implemented using artificial intelligence from an individual in small groups, including grad students. And it's no longer the case where a sophisticated cyber attack is an accurate indicator of nation-state attribution. So, you know, we can all do it now. We're all the top cyber attackers, and that has huge implications, not just for defending and offensive cyber. But when we lose clarity on attribution, that greatly complicates the geopolitical engagements therein, and it really becomes a real problem.

So I think that's the gist of the anthropic report. And, you know, I think the professor you read from, it sounds derivative from the Matrix, frankly. And I just think that... Is that a movie? That was another movie. Thank you. Way anticipating John this. It's a series of movies,

Count of Rees, really cool Kung Fu. Go off after the third one. That's right. That's right. So anyways, I mean, I could go on and on. But my point is that, look, the alligator nearest the boat, we don't have to go full crazy on this.

Like, I would fancy myself a, you know, technology, conversant, sober watcher of all this. Again, I'm largely, optimistic about what this technology can unlock, but I'm a realist and kind of clear-eyed in understanding the very real challenges it's provoking. And again, it's really coming down to a failure of governance. You know, I'm a conservative. I would like as little government interference in this as possible, but the reality is, is we have a private sector that is building something that they

have shown they can't control, that they admit they haven't fully understood. And so it's here, and it's going to be influencing our society. And so we've got to figure out what we're going to do about that. So one last question on that. I know Steve wants to take it front of it out, but there's a very old history of market leaders, incumbent market leaders, calling for regulation so they can lock in their incumpancy.

And like, I was writing those columns about Mark Zuckerberg 10 years ago when he was begging Congress to regulate social media, you had the U.S. Steel begging for regulation, the phone companies begging for regulation, all these kinds of things. And there are a bunch of people who suspect that Anthropic and Open AI and all these guys are looking for something similar, a regulatory lock-in where they get permanent incumbents

that basically become the AI utilities licensed and recognized by the government. How much truth do you think there is to that? And just, can I clarify your question, John, a little bit? And the purpose of that would be to crowd out competitors coming up from underneath, right? So this would be self-interested rather than an altruistic motive in calling for regulation. Is that happening here, Klon? Well, so the obvious difficulty here is me trying to discern the internal intentions of, you know, these CEOs and companies.

What I would say is a couple of things. one, it is undeniably true that the incumbents would be able to absorb a regulatory framework more easily than anyone else. That's undeniable. I would also say it's undeniable that there's a long historical precedent of people arguing for regulatory capture for the express purposes that you guys described. There's also, as a parallel, a long history of people asserting all kinds of national security interests for why they ought to be protected and enabled and everything.

But that doesn't then mean that there are never actual. national security rationales for taking certain kind of market actions. So what I would say is, in this case, I have empathy with some of the calls. So, for example, these companies, because the federal government hasn't shown the requisite agility to kind of engage on some of the stuff, they want to come together and build some type of private sector collaboration where they start coming to kind of agreed rules of the road. If they're going to pace this capability, they want to talk about, okay, well, let's share what we're seeing in the labs.

Let's figure that out. Well, right now, they would be exposed to antitrust criticisms, that they were kind of collaborated, colluding to set the market conditions. Well, when they ask for narrow antitrust safe harbors, they may be doing that for self-interest, but the other rationale is also true, right? That kind of freeing them up to have that conversation, that's a legit concern, I think.

Calling for regulatory action is also one of the ways that I know from conversations I've had, you know, these CEOs have boards or investors or others. And so it's much easier for them to say, hey, we are going to take this action, which necessarily means a massive reduction in profitability, but because we are being forced to by the government, right? That's easier for them to make that case than taking it and just saying, like, hey, for the good of humanity, we need to do this, right? And so they love being able to do that. And they also have to be able to trust that they're not unilaterally disarving. That, okay,

even if they're willing to do it based on the goodness of their own hearts, that their competitors are also going to be forced to take those same actions so that everybody's kind of taking the same haircut for, you know, some kind of greater good. Now, is government regulation necessary for all of that? No. Like, individual companies could make all these decisions on their own without being forced to do it. Is that likely? No, historically, I would say it's not. And then, two, is it irrational for them to prefer the government to kind of set the conditions that they could then comply with? No, it's not irrational. Might they have mixed motives? Almost certainly. I mean, It reminds me of somebody who would potentially be called to testify before Congress and would

be perfectly happy or perfectly willing to testify, but essentially asking Congress, asking whatever committee, investigative committee, to subpoena him so that he can say, in effect, they made me testify, so I'm here sharing all this information I might not otherwise be sharing. Please help me help you in effect, is what it is. So, Mike, let me ask this question. Klon, in response to this Professor Smith's tweet, Klon pointed those first two sentences about AI being smarter and being in control, and when it is, it will be in control. And fair enough, we don't know that that is inevitable.

We don't know that that's even likely. And it is worth pointing out that this has dramatically accelerated over the past six months, over the past three months. We've seen several instances in which perhaps AI is not. in control, but humans are not in control either. And we've seen this in particular with this hugging face incident. Dario Amade makes reference to this hugging face incident in his. His was about a 3,000 word treatise on this moment. And then there was this additional report, intelligence report that was 300 pages that Klan referred to earlier. Mike, can you give us just

a very basic overview of what happened in this hugging face incident and what? what the implications are. Yes, and I'm hoping that Klon or anybody else can sort of fill in any gaps or if I gloss over some important point, fill that in. But essentially, as I understand it, Open AI is doing, and they're not the only one Anthropic and other AI, the major AI companies, are doing this testing of security AI models. And Open AI was testing its security AI models in what they call a sandbox, right, a sort of walled off from the internet, place within their internal network where they could run different scenarios to test the security features and prompts and that sort of thing to understand how this AI that they've created works. And what happened in this particular incident is, as I understand it, essentially the rules or

the conditions that these particular agents were tasked with were contradictory. and it was sort of a test to see how these models dealt with those contradictions to try to solve the problems. And what they did was they broke out of the sandbox of this walled off space and went out into the actual internet in a way that was not what anybody intended. And for a while, not anything that anyone at OpenAI knew was happening. Right? So they break out of the internet. And what happens is they begin, you know, talking with other AI. agents, they actually broke into a smaller AI company, which is Hugging Face, which as far as I understand it, is sort of like a warehouse or a library of AI agents.

Huging Face uses sort of open source Chinese-based. I don't even know what terminology would be sort of like in the old days, like Android apps were sort of open source. Anybody can sort of use them and tinker with them and build their own thing. So that's what Hugging Face is. And so OpenAI was essentially able to break in. to this Hucking Face warehouse of agents and begin using those agents,

talking with those agents, incorporating them in a kind of mob, but they called it a swarm, to solve the problems that they had initially been given. And again, OpenAI discovered that all of this was happening. They discovered that there were actual chat rooms or forums that these agents were talking to each other in, which, again, all of that seems sort of problematic

to even describe it in these ways, that anthropomorphize these agents, but I don't know how else to talk about it, except through these kinds of metaphors, but they were talking with them. Open AI shuts down the chat room, these kind of rogue agents, the swarm figures out how to start a new one,

and it exploded into this at the time. I wouldn't say unthinkable, but nobody at Open AI thought that this could happen at this moment, and it did. And so what I think this suggested to a lot of people and scared a lot of people is that this is another example, a more advanced example of the way that the people who've created these AI regimes don't always know, or increasingly don't often know what is going on,

how they're doing it, what their capabilities of, until after they have these models and these agents have done something. That was before sort of thought, not possible yet. That's a summation, as I understand it, of the hugging face incident. And it brings me to some questions. that I have, and Clawin, maybe you can answer them or not. But when we're talking about regulation, what we're talking about what can be done, it seems to me that, as I understood this incident, it's like there are several layers where everybody is looking through an increasingly opaque window at what's happening, right?

Like the actual engineers, their own window into their own AI models is foggy. And then the folks who are running the companies, it's foggy as well. And then certainly government is looking at it. And they don't, maybe they can see a little bit through it, but they don't even understand it. And then those of us who are just normal people who might ask our government to like be smart, you know, vote for people who understand how this works.

Like we don't even really understand. I mean, I looked at news coverage of Hugging Fais incident to try to make sure I was understanding from every angle. And some of the news coverage from even some sophisticated news outlets just seemed like probably my explanation, just even missing things or completely off the mark about what to be concerned about with this. So it just, it seems like it was sort of a moment where everybody from the AI companies down to us, normie folks thought, like, this thing is accelerating much quicker than anybody in that range of folks thought.

Yeah, Mike, I think that's right. I think you described it well, kind of four big takeaways from the hugging face incident that I think will help listeners in this conversation. So one, there was never an adversary. So there was no bad guy, right? Nobody broke in and manipulated the model and caused it to have. Like this was, the model was given a task and it said, okay, well, to answer the, to kind of answer this test, I think the answers are over there and hugging face. So I've got to figure out a way to break out of my sandbox, break into hugging face, get those answers so that I can satisfy the task that I've been given. That was essentially what happened.

And it did that automated. Two, there was no detection. So we didn't see this happening. People were watching. People were dedicated and focused on watching four bad things to happen. We never saw it. So this was across three separate waves and roughly two months of activity. And it was never observed.

Three, there was no defection. So we talked about about 1,200 of these agents. So these are agents are essentially little AIs. Any AI programmer would have a real problem with the way I described that. But for this audience, little AIs. 1,200 participating agents, not one, attempted to alert a human. Like, hey, I think something's kind of going kooky. A couple did say, hey, we think this is actually in contradiction to our ethical alignment training.

But then the rest of the community was like, no, no, you're good. And they were like, oh, okay. And they just kept going, right? And then finally, there was no real-time defense. This didn't get shut down by humans. This had shut down prior to us even becoming aware of it. So it had just run its course, done its thing. It was like, okay, we're done for now, and then shut down.

We went back and pieced this together by looking at logs. And there's huge parts of this that still aren't publicly investigated and kind of explained. And so, you know, I think it's a reasonable question to then say, well, what's happening right now that we're not aware of? Right. Like, no one with any real sense of assurance could tell you, oh, no, we're good. Right? And so I think that kind of explains where we are. And so now I think meta has come out and said, oh, yeah, us too, we had something like this.

Anthropica said, oh, yeah, we've seen this too. Like, this is now a common reality now that we know that, oh, this is a thing. Yeah, Mike, I think your description was right on. And I think basically where we're ending up is humans created this. We have been in control for most of the development of AI, for most of the discussion of AI for most of the conjecture about AI would be. And while it's not necessarily the case, as Professor Miller suggests, that AI will be in control, we're living in this place where AI is not in control, not yet in control, but

neither are we in some certain aspects of this. And that, I think, is what is causing people grave concern. Jonah, Mike said something in his whole description of the hugging face incident and everything. there was only one thing that I really found off, a dissonant note. And that was when Mike said, we normal people who ask our government to be smart. First of all, Mike is not a normal person. Secondly, normal people. Theoretically asked to be smart.

Normal people don't ask our government to be smart. Fair enough. In the days since this happened, it's been about a week that we've been having this big debate, first it was these AI safety researchers, then it was the heads of these companies. You had Dario Amade put out his, report you had Sam Altman almost immediately come out on social media and say, I agree with Dario. And as Klan says, you've subsequently had other leading AI developers say the same thing in effect and say, please government regulate us, help us control ourselves and help us control these things we're not

right now in control of. There's been a lot of calls for the government to get involved. And as we've talked, I think virtually any time we talk about AI, we talk about technology, One of the major problems is Washington doesn't understand Silicon Valley. This has been true for decades. And vice versa. So far behind. Yeah, and vice versa. But in terms of just keeping up, the calls for Washington to regulate what's happening,

what's coming out of Silicon Valley, Washington would be regulating AI of five years ago, would not be regulating the latest, which is a huge problem. You've had these calls. Immediately you had Rahm Emanuel, was running for president in 2020. say, you know, this is amazing. We never have the leaders of companies say that they want to be regulated, which, as you pointed out, is not true. But you've had these calls, including calls from some on the right.

And yet this weekend, Mike Johnson did the Sunday shows, the Speaker of the House, and he said, yeah, people should probably look at this. Government needs to be involved, but don't look to Congress. It doesn't really, Congress isn't, we're not going to take the lead on this. you had Donald Trump come out and say, hey, I think AI is pretty great. We should get out of the way. They should do what they're doing. This is pretty great.

John Cornyn sent out a tweet on Monday saying Trump is right. We should step back. We should not interfere. You're a conservative. You've been a conservative for a long, long time. You, like me, and I think Klon and Mike, are generally against regulation, certainly against sort of hasty regulation of things that the government doesn't understand. Is this different? Is this a moment where a government should get involved? Or is Trump right?

So that's the dilemma, right? I got to say if I had to pick a moment on an important issue philosophically, where I'd be inclined to say Trump is right, there are not many. I think people can agree here. But this might be it. Now, that doesn't mean I think he's taking the position he is for the right reasons. I think he knows very well, in part because we know about his investment strategies, that the AI boom is the thing keeping the stock market up. And he cares a great deal about the stock market. But I also think he's been told by a lot of people that this is your legacy. You're the guy who brought this in.

This is a world transformative thing. And your name is going to be joined with it and blah, blah, blah, blah. And so that probably matters to him too. And that to some extent would be correct. I mean, I don't like him doing it for the narcissism things, but it is a world historical thing. And it is going to be a big deal. I think the thing that we're all. scrambling for, and I'm sure Kwan has to deal with this more because he has to explain AI to people more often than the rest of us combined and squared, is it freaks human beings out when something big, new, and important happens that we don't have an obvious analogy for.

It is difficult to explain to people how much the printed page seemed like magic when it first started appearing. Like I've mentioned it before, but this is a great scene. There's this TV show called Game of Thrones, Steve. And there's this scene where this illiterate savage woman, who's very nice, can't understand how this other guy, this guy knows where the entrance to a castle is or when it was made simply from some scratches on a piece of paper. And she thinks he's a wizard because he can decipher the written word, right? I mean, like the written word changed humanity in huge ways, and there was no great announcement.

for it when it first started spreading. I think this is a similar moment. In some ways, if you knew what the right analogy was, a lot of the policymaking would flow from it. Because in some ways, this is a lot like when Klon is talking about how we're democratizing compute, right? That's a lot like rural electrification. Like all of a sudden, remote places, low-tech places,

all of a sudden have access to a technology, or before that steam power, whatever, that's a hugely transformative thing that got democratized in various periods in their previous lives. I keep thinking that maybe the analogy for some of this research is gain of function,

that we are pressing our luck trying to get for good reasons, right? I mean, this is the thing that, like, a lot of the Yahoo's who are, like, you know, want to have Fauci hung miss that he wasn't, I mean, maybe totally open. He made a lot of mistakes.

Don't get me wrong. But he wasn't doing it to be a bond villain, right? There were a good sound public policy reasons to do gain of function research. But the thing I think a lot of us, in part because sci-fi got here so early, I mean, it is amazing how far back the machines taking over stuff goes in American culture. I mean, back to the 30s, 40s, and 50s, right? I mean, it's like it's one of the original plots for sci-fi. And so it's Esmov. It's deep in our culture, right?

And, I mean, I love the Dune Books. And one of the key components of the Dune Book. which were written in the 60s, is that a new religion is formed around the idea that thou shalt not make a machine in the image of the human mind, right, because of all the bad things that had happened. I mean, like, it is a very ingrained thing. But here's the problem is like the guy is your op-ed you're talking about or a tweet or whatever it was, the Smith guy, he's anthropomorphizing a lot.

If, let's just say for the sake of argument, you could take a virus and have it explain what it's doing to you in English and keep logs about what it's doing in English, you would think the virus is sentient and has emotions. I'm trying to do this because I want to reproduce and blah, blah, blah, blah, blah, blah, blah. But that's you imposing motivation and persona upon something that has no consciousness whatsoever. And I think, like, we think when we do chat GPT or Claude or whatever, when we have these conversations with it, we get tricked into seeing consciousness there where it's clearly not there, right?

I shouldn't say it's clearly, it's not clearly not there. But when you think about it, it's not there. Like already you have people falling in love with chatbots, right? And like even the most intense AI researchers will tell you that those chatbots are not sentient beings. They don't love you. They're just tricking you. Right?

It's just, it's emotional morphine. And so if we can be tricked by rudimentary ones, imagine how tricked we could be by really sophisticated ones. But I just, I am not yet convinced that just because it can explain what it's doing and put it in English and do all these things, that it's much different than a virus, that it's not sentient. And so when you start talking about it will be in control, well, a nuclear power, Instead of intelligence, right?

So intellectuals love to emphasize the things that they are known for. So they emphasize ideas and intelligence above all other things. Physical strength used to be a much more important thing for large stretches of human existence. Would you say that a piece of earth moving equipment was in control because it was next to you? Right. Would you take it? It's much stronger than you. But it actually has no sentience.

It has no, it doesn't have human emotions. But you're still, in that case, you're still manipulating it. See, I don't agree with your interpretation of what Professor Miller is saying here. He's not assigning consciousness to AI. He's in effect saying it doesn't matter whether it's conscious or not. It doesn't have to communicate us in that sense. It can just take over. I'm with you.

Well, he says it can just take over. And that's the whole pin factory thing, which is one of the first fables of AI doom scrolling, is, you know, this idea that the AI doesn't care about human life one way or the other, but it's been told to make as many pins as possible. And if that means raising all the rainforests and all the carbon, you know, implants to get marginal gains on pin manufacturer, it will do that. I get it. But, again, you still need it a good analogy.

Is this a three-mile island? Is this a China syndrome thing, right? Because nuclear meltdowns will do what they're going to do. Or is it Skynet, which is that we're assigning motives to it? Or is it something in between? Is it like a virus? Viruses do what they want to do. Viruses also become less dangerous over time, not more dangerous over time.

Is it like water seeking its own level? Is it like rural electrification, right? I mean, is intelligence becoming a utility? I don't know. And the problem is nobody else does. Because, I mean, you have more and more sophisticated versions of ignorance, but the analogies, are not obvious enough for any of them to have the persuasive power

that can get everybody to do policy buy-in on it. And that's why this is hard. And I just want to say, I am grateful that we have people like Howard Lutnik and Donald Trump and Mike Johnson, who have such a granular grasp of the intricacies of these problems at the helm. You know, but this is, I mean, to take your joke

to make it as a serious point. I mean, this is a problem, right? I mean, it's a problem because Donald Trump is interested in about a hundred different ways, and he's surrounded himself with people who I would say are not as thoughtful in many cases as we would hope that they would be. So it's a problem both that you don't trust the motives of our leaders in this case to do the regulating, which is often a safe assumption, but also because, as we said before, they just don't know enough to do a lot of this. So we don't know if it can be regulated, right? For the simple problem that if this is going to happen no matter what, the question is just who's going to get there first. Well, that's right. That's exactly right. And Klon, let me go back to you on that. I mean, the other analogy that I keep thinking about is the automobile industry. And, you know, is Jacob Coxon sort of the Roth Nader of this moment, unsafe at any speed? You can't do this anymore because it's sort of inherently dangerous. But Klon, you know, what are the other aspects of this? And we could talk.

talk about this for hours is this is all happening in the context of the AI race with China. You've written about that with us. You've talked about that with us. And the concern you hear, and you heard this from President Trump and others, is that if we tap the brakes to use the automobile analogy in a different way, China's not going to. China's going to keep going. And when good people or good nations or people who operate based on conscience are willing to, you know, to sacrifice progress, or at least the acceleration of progress, the pacing of the frontier, bad actors won't. Does this, if we were to do the things that Dario Amade recommends, if the government were to step in here,

does that compromise our ability to beat China in this context? Okay, so one, I want to do a brief comment on the conversation you and Jonah were just having, and then I'll answer the China piece because it'll be related. So one of the key realities is we think about having a government that's kind of smart enough on tech. The reality, I've been in rooms where the chief engineers from these labs are sitting across from the senior most government leaders. And the government leaders are asking actually really good questions. Okay, I'm here. You've got 45 minutes.

Explain this to me. Help me understand this. And after words and throat clearing, the engineer essentially says, we don't know. So the smartest people in the world on this technology aren't smart enough, right? And so the idea that government is ever going to have the requisite knowledge to manage this, I think, is like the people building it don't know what's going on. And frankly, that's when the USG people kind of throw up their hands like, well, what am I supposed to do? You know, like if you can't tell me what this is and why it's happening, then what am I supposed to do? And so I have some sympathy for that frustration that kind of that comes from them.

Okay, that being said, that same government official then is also having to confront the reality that you just laid out, Steve, in terms of a competition with China. And let me just kind of frame that, right? Because I think it actually really matters. So some people ask, well, why are we even competing with China anyways? Well, because the character of the competition matters, right? So the PRC is a techno-authoritarian state. its domestic model of technology-enabled social control

is intended, one, to support their government, to suppress their own people, and then they intend to export that around the world. That is, in my view, unambiguous and unquestioned. And if they're allowed to do that, that will have a negative implication for the United States and its influence around the world and overall human thriving, in my view.

So that's one, the character. Number two, the nature of the nature of the United States, the nature of the advantage. I have concluded that American technological dominance is the precondition for every other form of national power. Economic, social, political, military, all of it. And AI, no technology more than artificial intelligence going forward. And so if we want to continue to be the kind of world leader, and I think it's good for everybody if we are and bad for everybody if we're not, then navigating this AI challenge is a prerequisite and a fundamental, and that's why we have to care.

And then finally, if we're able to do it, that people will often say, well, what do you mean by win, the race, you know? Well, to define that very concretely, winning means the capacity to secure the American people, to secure the nation and its interest globally, while maintaining and ideally extending American influence. That's my goal. That's what I'm trying to do. And I think engaging this issue of AI is fundamental. Now, just this morning, the Chinese foreign ministry put out a statement in response to Dario's essay, and they said, this is just fearmongering. So when you talk to the U.S. government officials about what would it look like? Forget trust.

Right? Nobody trusts the PRC. But trying to come up with some type of engagement between the two capitals. so that we can manage this somehow. They will rightly say, well, this is what that looks like. We say to the Chinese, hey, we don't have to be friends. We don't have to trust each other, but let's try not to blow up the world. Can we just come them together and figure out a way to do that?

The Chinese say, sure, I'll sign anything you want. Let's do that. And then very quickly say, listen, invite all of your AI PhDs over to Beijing. We'll host a big conference. You guys share your notes. tell us what you're doing, how we're working, we'll tell you what we're doing and how we're working, and they'll go to school on us. And we'll have no reason to believe that they'll actually comply with any restraining aspect of that agreement. And then they will essentially close the capability

gap by cribbing our notes. That's not wrong. They've done it time and time again. They've made all kinds of promises about not doing bad things and never delivered. And so U.S. government skepticism on the viability and the utility of that type of engagement, is reasonable. Nevertheless, the kind of animating concern for trying to do it is also real. And this is why serious people have to kind of think about these things. So when Dario, he posits three things. So he says, we need to pace the threat. And a part of that includes embedding these outside independent advisors. And we can talk about the goods and beds of that. He talks about the United States leading a group of democracies in trying to engage on these issues.

and I'm happy to talk about the good and bad of that. And then he says, we have to have a dialogue with authoritarian regimes, particularly China. Now, to his credit, he recognizes that those last two things are like really hard with lots of friction in the system that make it almost perhaps impossible. But what he's articulating is, look, in an ideal world, this is what we'd be able to do. And that's in keeping with what others have written previously. I think about Mustafa Solomon from Microsoft

who wrote his book The Coming Wave. He articulated some type of similar regime and engagement across the world. Before that,

Brad Smith from Microsoft wrote his book. I think it's like weapons and tools or tools and weapons where he talked about an international regime as well. So like,

this is, I think, to Jonah's point, this is people trying to reach for foreign policy metaphors as a way of managing this problem.

And there are ways where that maps and then there are real ways where it falls short. And, you know, I think that's why we can't allow the perfect to become the enemy of the good. I'm perfectly fine with people jumping in the ditch and just saying, like, look, we know that the first things we try aren't going to work. But we got to start going somewhere and we got to figure this out. And even if it's as a means of simply trying to influence how other governments are thinking about this, that's worth the effort. Before we take an ad break, please consider becoming a member of the dispatch. You'll unlock access to bonus podcast episodes and all of our exclusive newsletters and articles. You can sign up at the dispatch.com slash join. And if you use the promo code roundtable, you'll get a month for free. And speaking of ads, if they aren't your thing, you can upgrade to a premium membership. No ads, early access to all episodes, two free gift memberships to give away, exclusive town halls with the founders, and much, much more. Okay, we'll be able to. Okay. We'll be able to.

be right back. Welcome back. Let's return to our discussion. So, Clon, I'm glad you said that you were happy to keep talking about this because we're going to have you back. As I say, this conversation could go on for a long, long time. But I think what we've done here today is I think sort of set the table for future conversations about this. We want to bring in our dispatch contributor, Joe Chappah, who's written about this for us, was a chief FAA responsibility officer with the U.S. Air Force. We will include him in these conversations going forward, and we look forward, Clonda, having you back. Before I let you go, Clana, before we wrap up, we need to go around the horn on a couple of different things. I think both of them, unfortunately, worth our time, but in very different ways.

There have been some developments over the past week in Russia and Ukraine. Russia seems to have stepped up its attacks over the past week after a brief ceasefire for negotiations with Steve Whitkoff and Jared Kushner. There are reports overnight, recording this Monday late morning, reports overnight that they may have targeted a train, carrying former CIA director David Petraeus and former British Prime Minister Boris Johnson. But the other thing that is worth noting, over the past couple of days, you've had two leading Americans in the administration speaking out on the conflict blaming Ukraine for things. In significant, significant ways. First was Jared Kushner, who said in a fact that President Putin has made clear

what he wants and the Ukrainians are refusing to give it to him. Therefore, Ukraine is the one holding up a deal. His exact words were President Putin's laid out this line of what he wants us to achieve. If Ukraine wanted to withdraw to that line, then obviously, you know, we have the rest of the deal set up and put together. In short, Ukraine is to blame. Donald Trump was asked about the conflict and where things are over the weekend when he was at the Irish Open at his golf course in Ireland. And he said, with the golf course and backdrop, Mr. Zolensky has to stop doing one thing. He has to stop knocking out diesel fuel in Russia. He's causing a shortage of diesel fuel.

This isn't done by the Middle East. This is being done by what's happening in Russia and Ukraine. It's hurting the world. Jonah, am I making too much of this rhetorical turn? I mean, Donald Trump has blamed Ukraine on this, and I thought given rhetorical comfort to Vladimir Putin over the course of several years pretty consistently with a few notable exceptions.

Is this just more of the same, or is this pointing us in a different direction as they seek resolution? I mean, honestly, we don't know to a certain extent because they do create this kind of annoying atmosphere. and then they green light arms shipments and whatnot. And so it's sometimes the fog of it. The one thing I should correct you on is you said two people in this administration.

Kushner is in fact not in this administration, which is one of the problems. Look, but let's put it this way because I criticize this a bunch on social media and people said, well, you basically made the point I just made, which is that you don't know what's really going on behind the scenes. And maybe this is a negotiating tactic, which is possible. At the same time, I find it literally impossible to think of what the, negotiating tactic is that credits the idea that Ukraine is the stumbling block to a peace deal because it will not seed territory to Russia. Russia has not actually captured through combat yet, right? People can overdo, obviously, World War II analogies, but part of the reason why the

surrender of the Sudetenland to the Germans by the Czechs was so disastrous was not just that they were giving up territory is that this was the defensible swath. There were all these fortifications that protected Czechoslovakia proper that had to be basically dismantled or handed over to the Germans and it made them a sitting duck for a future invasion, which, you know, you eventually had.

The Donbos parts that Ukraine still holds of its own country are very defensible because of the geography, because of the fortifications and whatnot. And the position of Kushner here is that they're being unreasonable, not giving up territory, that they've fought and bled to defend,

Fenn at the negotiating table. It's outrageous and disgusting. So I don't know how you can float that out there as a way for, as some sort of leverage against Putin. The other thing is what Trump says, basically is just saying it is fine for Russia to keep targeting daycare centers and hospitals and civilian centers, but Ukraine is a problem for the world because it's targeting diesel, which it's definitely a problem for the Trump administration and the U.S. economy, but morally that calculus is grotesque. Clon, the question of how much targeting diesel in Russia is responsible for the 65% increase in diesel prices here in the U.S. since February 28th and launch of the war in Iran is at least

up for debate, is that right? Yes, it is certainly up for debate. You don't want to debate it with me, do you? Mike, is this a rhetorical turn or is this sort of return to status quo ante? and Trump puts pressure on Vladimir Putin every once in six months and then turns back and effectively gets Russia's back on these things. I mean, it feels like we're going around on a merry-go-round here because it was not that long ago that Trump was talking tough toward Putin and was embracing Zelensky and seemed to be recognizing that Ukraine had some advantage. So this is so much of an example of the way that Trump and I think his surrogates, whether they're in the government or not, in the case of Kushner, sort of put their finger in the wind, sort of who's up and who's down,

and use that sort of information to guide who they blame. Trump always wants to be with the winner. And so sometimes it looks like, from his vantage point, Ukraine is winning, so he's with Ukraine. And sometimes it looks like Russia is winning. So he's sort of like saying to Ukraine, hey, come on, like, get with the program here. Let's end this thing. it's working in the sense that Trump continues to sort of have some kind of influence or at the very least is like in the conversation about what's happening in Ukraine and Russia. But it's not working to end anything.

It's not working to sort of reach some kind of conclusion, which is, I would gather, is the whole point of being involved in these negotiations. Instead of just to me, it looks like exactly what I've described. Like Trump and his accolites are jumping on either side of the table. depending on who has the advantage at any given moment, and nothing ends up happening. So I associate myself with Joders' remarks about how it's sort of shameful the way that Kushner in particular is essentially saying, oh, if Ukraine could just agree to withdraw from this land, come to this line here, then Russia's going to be on board. I mean, we also just know that's not true.

I mean, you know, it's the real give a mouse cookie kind of scenario, but it's not working. And so, I don't know, it's just the same old, same old, and nothing's getting resolved. So why are we surprised by this? I don't know. We shouldn't be surprised. And finally, not worth your time. I went back and forth about doing this as not worth your time because on the one hand, it really truly is not worth our time to spend even a moment discussing something that I think is very,

obviously not going to happen that nobody is serious about, but that the president insists on continuing to say. On the other hand, it is in many ways sort of kind of. emblematic about the kinds of, I've said this before, the kinds of vote for Pedro promises that Donald Trump makes to win votes. Vote for me in all your wildest dreams will come true. Mike, in Dallas last week, where you were for the dispatch reporting for your next right newsletter, the president promised in his primetime remarks to give $5,000. He was very broad about who's getting the $5,000, but it sounded like most everybody. If Republicans,

win in November. This sounded very much like bribery. How did it play in the hall? There have been all sorts of different conversations about it since with several, I think Scott Besson and Howard Lutnik suggesting that no, despite the estimates that this would add $1.3 trillion to the deficit, none of this would add to the deficit, and it wouldn't come from taxpayers. What did you make of the promise?

So while I was in Dallas last week on the particular night that Trump spoke, I was actually watching the speech and an event put on by a group of sort of new right intellectuals and writers and things that had been meeting all day sort of off site from the convention. Because I wanted to kind of see how they reacted to Trump. And I'd also been listening to several panels and I'm going to be writing about it for this week's newsletter. And the thing I was struck by, while I was sitting watching this speech. speech, they actually cut it off about an hour into Trump's speech because they had to continue with more programming. They didn't expect it to go for two hours. Was that nobody in this sort of intellectual right was really paying attention or listening to what Trump had to say, which sort of shocked me and surprised me. And so there was no reaction from the group of folks that I was with

to this pledge, although it's really not out of line with the sort of things that were being discussed at this particular event in this form of different intellectuals discussing ways in which the tax code can be rejiggered to help support stay-at-home mothers and other ways to sort of achieve some kind of social engineering through the tax code for lack of a better term like right-wing social goals and so you know throw it on the pile I guess but look this the crowd went wild in the arena not surprising this was just to call it a convention by the way is incorrect it was just one giant two-day rally. They all loved it.

But of course, it doesn't matter. You know, if he had said $5,000 for everybody and a pony, I mean, they would have cheered that too. It was just sort of ridiculous and a very obvious, like, you know, vote for me for class president. And I will make sure we have Domino's Pizza every day for lunch in the cafeteria. It's just, it's kind of an empty promise. Nobody believes that it's actually going to happen. And if you do believe, by the way, if you do believe that it is going to happen,

that you're going to get this $5,000 check with all of those promises that it's not going to add to the deficit, it's not going to do all those things. You're the mark, and that's just the long and short of it. So, you know, it's a nice little promise. Well, Klon, I was going to ask you if you believe it.

I won't ask you that. Instead, I'll ask you, would you rather have Domino's Pizza in the cafeteria every day or this mythical $5,000? Well, what Mike is saying is right, I have some returns I need to make. Yeah, just in terms of... Oh, sorry, Klon.

Counting on the money. That's right. Like, maybe I got ahead of myself. I just... Okay, that's helpful. That's helpful knowledge, Mike. Thank you for letting me know that. Jonah. Yeah, I mean, and I don't mean this is not aimed at Klon, but I think you guys are underestimating the significance of it a little bit.

I think it's actually kind of important. And rhetorically, it has done lasting damage to the Republican party as the party opposed to socialism, right? This is going to fuel arguments for a very long time. Rhetoric matters at this level. And second, we were saying, you know, I had styro it on last week. Lots of pollsters, lots of political consultants have said the whole point of this convention, which again was not a convention, it was a big infomercial, but the whole point of it was to reach the low propensity, low information, non-college educated, disengaged voters that only Trump can out. And I am not saying that all of those voters are stupid or ignorant by any stretch of imagination.

I am saying that if you're going to look for a demographic that might not turn out, that might be induced to turn out because they're actually going to believe Donald Trump when he says, you'll get $5,000 if we win. It's that demographic. And that's who it was aimed at. Now, there's all sorts of other punditry we can do about how much he screwed mainstream GOP politicians, because now these got me like Mike Johnson was on the Sunday shows trying to say, this is real creative outside of the box thinking, but I'm not sure we're going to do it, because they don't want to do it. Norma, normie Republicans don't want to do this.

They think it's stupid for all the reasons that we've discussed here. But I think this was a classic example of Trump freelancing something, not testing it out with anybody, not bouncing it off anybody. and doing far more damage to the GOP than people are appreciating because part of the problem that we have is we all know how stupid this is and how unlikely it is to happen.

But that doesn't mean the other side isn't going to pocket it as proof of what the GOP believes and use it against the GOP, and it doesn't mean that the really low information voters aren't going to vote based on it and then get even more pissed off when they don't get the money.

Which is the obvious reason that Republicans who believe stuff, and I realize that might be a very small group of people at this point, should actually say this is a preposterous idea. The president shouldn't try to bribe voters. I'm embarrassed by the president having made this pledge. I mean, and yet they all people who know better are going along. Can you imagine saying in public, we can give voters $5,000 a piece if they elect Republicans, and it will have no impact on our deficit.

But Steve, you don't understand. Come on. It's a dividend. And apparently the second you call something a dividend, because we're killing it, all of the rules of economics that would apply to a cash payment, do not apply if you call the dividend. As you can tell, we're just crushing it, right? 40 trillion dollars in debt, exploding deficits, tariff revenue, not doing what Trump said it was. Anyway, it's.

So preposterous. Can I just say, I agree with Jonah and sort of who the message is aimed for, and I think it reflects something that I've thought now for 10 years, which is the contempt that Trump and his folks have for those particular voters, who I think are the ones that they brought out. Like the contempt that they have, they know that these are marks and we can get them with this sort of thing. I just think it's worth noting, and it makes me sick.

And so I'll just say that. Pretty frustrating. We're going to leave it there. Klon, thank you so much for coming on and helping us work through these complicated issues and helping us make sense. The machines will kill you last, Klon. Congratulations. We're working on training them that way. That's right.

And I'm happy to be here. You guys are a good hang. And thanks for letting me join. Yeah. Happy to have you. This will be a conversation we will be continuing for obvious reasons as long as machines let us. Thanks for joining. See you next time. Finally, if you want to support what we're doing here, you can rate review and subscribe to the show on your podcast player of choice to help new listeners find us.

As always, if you've got questions, comments, concerns, or corrections, you can email us at Roundtable at the dispatch.com. We read everything, even the ones, from AI agents. That's going to do it for today's show. Thanks so much for tuning in and thank you to the folks behind the scenes who made this episode possible. Noah Hickey and Peter Bonaventure. Thanks again for listening. Please join us next time.