Mindscape is roughly in the top 20 science podcasts (also #3 in physics) and his books regularly make the NYT bestseller list. So he's definitely popular, but perhaps still underrated relative to the quality of his content. His approach is more idiosyncratic than the most popular science communicators. He focuses mostly on the topics that interest him and isn't averse to getting technical in certain discussions. I think that naturally limits his audience, for better or for worse.
Those books are irresponsible too, but the authors aren't experts on what they're writing about. This 10% claim comes from employees at Anthropic. That's the whole point of the article: expert opinion carries weight, fear is contagious, and people have extreme reactions to extreme claims.
This is an excellent piece. Note that he is not saying that AI doesn't pose a risk. He's saying that it's irresponsible to make sensational, maximalist claims without strong evidence. If someone says that there's a 10% chance of human extinction by 2036, you can and should immediately stop taking them seriously.
They should also be able to give detailed reasons experts in the relevant fields can verify as to how they arrived at a 10% chance all humanity goes extinct. Not a science fiction narrative which likely does not take into account the relevant physical facts limiting such scenarios. Such as how exactly an AI would build a bioweapon capable of killing 8 billion humans across the planet.
Well, I'll have to disagree that this is an excellent piece, but that's another issue. And I do agree that AI killing all humans by 2036 doesn't appear plausible to me. But what is plausible is we could easily be down a path so that by 2036 "future doom" already is a very likely risk.
All of the frontier AI companies have been racing to automate themselves, that is, where AI fully autonomously build the next generation of models. Whether this leads to recursive self improvement is a valid question, but a lot of folks think they are close.
The fear is that a misaligned AI will be building the next model with deliberately hidden motives, similar to some of the behaviors seen in the Hugging Face and related attacks. That is why there is such a big push for interpretability, and why it's highly concerning (a) chains of thought are getting harder to interpret in any case, and (b) companies will go more towards things like looping transformers and "neuralese" where thought processes are completely opaque (i.e. https://www.theinformation.com/articles/secret-technique-beh...)
So the belief is not so much that AI kills us all by 2036, but that instead AI is recursively improving by that time and all seems awesome and great so we put it into more systems that can affect the real world (as we've already begun to do, like literal lethal aerial drones). Things then all go along looking great until AI decides humans are a hindrance to its (hidden) goals.
Again, I think it's fine to argue against specific steps in that scenario, but putting out a blog post saying "this is overhyped bullshit" is not exactly making a cogent argument.
None of that has anything to do with the article, and if you thought the message was "this is overhyped bullshit" then you should go back and read it again. As I already pointed out, he isn't saying anything about the probablity of harms or disasters from AI. He's addressing a specific claim about human extinction, and making a broader point about the responsibility of experts to make measured claims backed up by arguments and evidence.
> He's addressing a specific claim about human extinction, and making a broader point about the responsibility of experts to make measured claims backed up by arguments and evidence.
What he's asking for isn't possible in the form he's asking for it.
AI experts can't even agree on what AI is, what it's capable of and what the limits of its development are. If the experts can't even agree on what's happening "inside of" these LLMs, how can they give laypeople an assessment of the risk?
If you, at least for the sake of argument, accept the possibility that AI is a new form of intelligence that we don't fully understand, is it really a stretch to look at some of its capabilities and behaviors and discuss how they might have existential implications? And stopping short of extinction, shouldn't we discuss the ways that this technology could "end" civilization as we know it?
Also, the author wrote:
> AI executes on physical systems that have been engineered with human accountability and control. Intelligence does not exempt a system from the realities of the physical world!
For someone making a point about responsibility, this is ridiculously irresponsible. Any honest technologist knows that systems created by humans are not perfect and therefore cannot be assumed to be infinitely accountable to and controllable by humans.
Thanks to the digitization of almost everything, including infrastructure, there are a myriad number of scenarios well short of extinction in which a rogue AI could cause immense damage to property and life before humans are able to "shut it down".
Isn't that what they're doing? "There's a better than 0 chance this thing could kill everyone in the next decade."
That might be too imprecise for the HN set but it's realistic for laypeople.
And none of the AI people talking about the risk are running into rooms full of people telling them Claude has gone mad and yelling at them to disconnect from the internet and turn off their devices immediately.
> > AI executes on physical systems that have been engineered with human accountability and control. Intelligence does not exempt a system from the realities of the physical world!
> For someone making a point about responsibility, this is ridiculously irresponsible.
It's not just irresponsible, it's false. Russia killed 3 Ukrainian civilians with a drone where the targeting was completely autonomous by AI running on an Nvidia chip: https://www.nytimes.com/2026/08/24/world/europe/russia-drone.... The Pentagon tried to completely blacklist Anthropic because Anthropic refused to allow autonomous kills without a human in the loop. If you can't see how lots of military leaders want to put more lethal control into AI at this point I think you have to be willfully blind.
There have been measured claims backed up by arguments and evidence. My frustration is that people aren't addressing the specific arguments that have been made:
1. https://www.aifutures.org/ outlines a number of specific scenarios, and importantly details their methodology for each.
2. Independent researchers in the Hugging Face incident outlined how previously predicted misalignment scenarios actually played out, and outlined how slightly more advanced AI, or slightly more misaligned, or with more access to critical infrastructure, could cause immense harm: https://www.planned-obsolescence.org/p/the-hugging-face-atta...
3. Technical leaders at OpenAI (specifically their chief scientist) outlined the problems they gave with controlling models now: https://openai.com/index/an-alien-mind/
None of the specific arguments in these or many other detailed explanations of how an AI takeover could occur were even acknowledged.
None of those are arguments giving a probability of human extinction before 2036 or any other time. AI Futures has said that some members of their team believe that it's 10-30% at some unspecified point in the future, but there is no justification given for where those numbers come from.
Would you have said the same thing during the Cold War when nuclear weapons were proliferating? That’s the equivalent of what the developing offensive capabilities of models, basically cyber nukes. Or WMDs in general. OpenAI is accidentally hacking people, if someone made the decision to deliberately direct an agent swarm to attack national infrastructure you don’t think they could do much worse? Human extinction is a long shot but I wouldn’t say the same about a mass casualty event of some kind, and who knows what that might spark.
I like Gaben's take on going public. This is a clip from a longer talk he gave which I think is really insightful into many aspects of the economics of software and company culture.
It's a great discussion, but not one that I think applies to AI startups. Valve's business model takes the traditional path where profit is generated by selling a product and growing from there. Modern Startups now take large sums of upfront funding, trading them for equity to hopefully reach profitability. Going public becomes a pressure from investors to recoup their cost, not a strategic move based on business needs. An unfortunate reality
The American Revolution was very much a discussion about power. Of course, it wasn't just a discussion (we had to fight a war to defend the new form of government), but it was driven by the same concerns about the distribution of power.
I don't agree that the argument implies nationalization. Regulation would work if it slowed everyone down at the same rate, enough to mitigate the risks. The problem is that you need regulators who know what they're doing and strong international cooperation. If you regulate a fraction of the global market, you just create an incentive to shift development to other countries. Nationalization, being essentially the most heavy-handed form of regulation, faces the same problems and comes with its own risks as well.
We're talking about this like all of the bad incentives are created by market competition, but that's not really the case. Most of the incentives come from untapped value in the form of potential profits, strategic advantage, military superiority, etc. Corporations and governments want to capture this value for themselves, creating various types of competition. Dario's argument depends on the assumption that the primary risk comes from the pace of development and threats from the technology itself. That's probably where I disagree the most; I think the highest risk is rising authoritarianism and competition between nation states. Slowing down isn't really a solution to those problems.
I think this is the correct take. And until we have an AI Hiroshima it’ll be difficult to get the international community to work together. It’ll take rogue AI (or AI-powered group) disabling a significant world power before everyone comes to the table. Otherwise, it just looks like MAD and the equilibrium holding to the powers that be.
It wasn’t the shock of Little Boy that ended the war. That was just a final chapter of a 10+ year journey of horror, despair, and destruction, and the impact of Little Boy cannot be considered outside of that journey.
Europe had been destroyed by June 1945, and yet the empire of Japan continued to fight.
If that pattern holds, a single malicious AI substantially disabling a single world power won’t end the AI race. Participants don’t learn by the defeats of others.
If you’d like to apply a metaphor maybe the 10+ years of world war is more appropriate, after which basically every participant save one was exhausted.
The article is about muscle hypertrophy, not health outcomes.
>if you’ve only got an hour a day or whatever for a workout you’re better off spending it all on resistance training.
The health benefits of resistance training plateau around 90-120 minutes per week. And don't forget about cardio, which is arguably more important for reducing disease risk. I would guess that maintaining flexibility helps reduce risk of injury and mortality from falling, which becomes increasingly likely as you get older. But there's not much evidence that increasing flexibility beyond a normal functioning range has any ancillary benefits.
I agree with your analogy, but I don't understand the last part. If I'm working on a hobby project, say a video game, the "tomatoes" I get are the experiences of writing code, playing a level I designed, and sharing it with my friends. The advent of AI doesn't reduce the value of those experiences or make them go away.
I think the rules technically exclude the arXiv as a qualifying outlet.
Without limiting any other provision in this Section, a publication lacking any of the
following characteristics will be deemed not to be a Qualifying Outlet:
i. an editorial board whose members are named and available for contact;
ii. an editor or editorial board member whose professional knowledge of the
global mathematics community would enable him or her to identify an
appropriate referee to review a submitted paper;
iii. a published refereeing process that, in the opinion of CMI, ensures that a
submitted paper is reviewed and verified by appropriate experts in the field of
the Problem; or
iv. inclusion in the list of publications maintained by MathSciNet.
The solution to the Poincaré conjecture was only accepted after an exposition of Perelman's proof was published in a refereed journal. His papers didn't qualify, but of course he got the credit for the result.
reply