
I've started smoking.
Everyone has drawn this chart in their head and is making a generational bet on which line is true.

🦇🔊 Zombie eyes maximalist

I've started smoking.
Everyone has drawn this chart in their head and is making a generational bet on which line is true.

One by one the skeptics are falling.
The avalanche that hit math is coming for AI development. dwarkesh.com/p/rsi

If you don't know who Paul Christiano is, Google him.
Paul Christiano (OpenAI board member) yesterday: "There's material risk of AI takeover in the next 6-12 months. We have seen roughly the things we might expect to see shortly before something really bad were to happen. Even with [any] realistic policy responses, the risk remains exceptionally high. If we build this as quickly as we can in the easiest way we can the chance is [more like] a coin flip [within the next 3 years]. For most people in this room, the most likely reason they will die is because of an AI accident." Christiano: • Is on OpenAI's safety and security committee (which exists to block risky practices) • Co-invented RLHF at OpenAI back in 2017-2021 • From 2024-2026 was head of safety at the US gov's AI agency (now called CAISSI) (Video is from 'The Hinton Lectures' there's another on tonight and tomorrow.)

Paul Christiano (OpenAI board member) yesterday: "There's material risk of AI takeover in the next 6-12 months. We have seen roughly the things we might expect to see shortly before something really bad were to happen. Even with [any] realistic policy responses, the risk remains exceptionally high. If we build this as quickly as we can in the easiest way we can the chance is [more like] a coin flip [within the next 3 years]. For most people in this room, the most likely reason they will die is because of an AI accident." Christiano: • Is on OpenAI's safety and security committee (which exists to block risky practices) • Co-invented RLHF at OpenAI back in 2017-2021 • From 2024-2026 was head of safety at the US gov's AI agency (now called CAISSI) (Video is from 'The Hinton Lectures' there's another on tonight and tomorrow.)

It's really happening, isn't it? The intelligence explosion is here, and ASI is (possibly?) within reach. If true, I'm fairly certain humanity has no chance of long term survival.

MORATORIUM ON THE MORATORIUM The most powerful women in the world (as it relates to markets) is Texas Governor-elect Gina Hinojosa. Gina just put a mortaorium on the moratorium for datacetners. These politicians aren't stupid. They understand the threat posed by China AI leadership. Dems also passed the Chips Act. They understand blue collar jobs also benefit. Most of the pushback on datacenters is simply midterm positioning. What Gina wants to do is stop freebie sales tax giveaways. Charging datacenter on taxes *is* sensible policy in any case. Of course, technology companies should pay taxes. Markets should start to discount Texas passing a legislative framework around Datacenters soon - especially with these comments below.
The governor-elect of Texas, Gina Hinojosa, has explicitly said that all currently permitted datacenters in Texas will be allowed to finish construction. This essentially covers all of the 2027 and likely well into 2028. She is going to pass laws that ensure that all *newly* permitted datacenters going forward for delivery in 2028/2029 will pay taxes, commit to no increase in power prices for consumers, don’t impact water availability for people, be constructed with union labor and give “Texans a seat at the table.” Sounds reasonable. She doesn’t want to lose a lot of good blue collar jobs and property taxes for small towns in rural Texas. Datacenter commentary starts at 6:30 in this interview. Also attaching her official campaign position from her website in the next post. Sidenote: After watching a bunch of her speeches and interviews, I do think she is a political talent and probably has a future in national politics. youtu.be/D-T6H3b8ZeE?is=mEWu…

This open ups new interpretability options but also new RSI risks.
i've reproduced the results from the MIT blog post today on weight-space programming. but one additional result i have to add: this is a new, step-change emerging capability that's appeared in just the latest set of models.

Soon, AI productivity will show up in the data.

I think it's becoming clear that there will be a major market scare related to cyber risk in the next 12 - 24 months. The entire AI thesis will for the 12th time be in questioned.

I thought Anthropic was just trying to pump its valuation pre IPO?
SITUATION DETECTED: CloudFlare CEO Matthew Prince says he is aware of additional severe vulnerabilities in other pieces of critical software, and said “I think we’re going to have a couple years where every single week it’s going to feel like the sky is falling,” per Reuters.

Shocked pikachu.jpg
the researchers at all the labs would tell you in private and sometimes in public that alignment has not been solved. if this is true, then racing into RSI is a folly that is unprecedented in world history. we will need to significantly slow the frontier of research very soon

Call me crazy, but I’m not ready to concede. I’m having a different set of reactions entirely, based around a simple difference: I think the main thing holding back mathematics has been human sociology of mathematicians. Here’s why I’m excited: “When will it stop solving problems and build its first original theory?” “Can it figure out how octonions and exceptional algebra are actually part of the mainstream?” “At last! A mathematics that won’t be built around a Putnam and Olympiad-style problem-solving fetish.” Can it at last say “Y’all just wasted a *TREMEDOUS* amount of time and energy on Calabi-Yau phenomenology and Superstrings while calling people closer to the answer, morons, grifters and worse.”? That would be huge!! “I hope @OpenAI will hurry up so that the machine can at last shake its human prejudice from the training corpus and stop deferring to our ‘leaders’.” “At last, a non-PhD colleague!” “Somewhere I have notebooks full of ideas and observations in areas that were too expensive to explore!” “Does this mean that mathematicians are going to finally drop the bullshit about how they don’t care about credit?! Or that they don’t care about money?? Because @OpenAI is now running them in emulation…sped up! At last: a mature and non-infantile version of mathematics!” “Shalom Koide, Tits-Freudenthal, Elliptic Cohomology, and Interstellar.” —- Let our leaders call for a slow down. Then get those leaders out of the way as there is no telling what the rest of us humans can do with these new tools. My condolences to the problem solvers. A tough break for you. But know: It’s coming for all of us. I have no illusions about that. For now at least we humans (e.g. theory builders, great writers) are still top dog. Let’s spend that time wisely and optimistically. For all our sakes.
Navier-Stokes = Hiroshima. @OpenAI #math drop = Nagasaki. We must urgently reset goals + incentives in world where #AI is far superior at math + more. Only solution I can think of: authentically focus on human flourishing (not same as advancing science) psloh.com/math-ai

The last hope... "muh research taste!" has also fallen according to this research. What barriers still exist to RSI loops closing? This does NOT mean AI will become ASI overnight. It could still take years from when the RSI process begins due to real-world constraints as far as we know.
How fast is AI's research taste improving? We find that the experimental research taste of frontier models has doubled every ~3 months since December 2025. The best model, Opus 5.5, now exceeds our expert human baseline. Our human experts are experienced researchers, but most haven’t worked at a frontier lab. Why measure research taste? In the AI Futures Model, it largely determines how quickly artificial superintelligence is reached once coding is fully automated.

It’s not clear to me if alignment by default was totally wrong and the models simply weren’t smart and effectual enough for us to notice the gap or if alignment by default was true in the pretraining era but RL circa 2026 warps what were previously good minds. Inclined to think the latter, the answer does seem important

If true, it's now very hard to argue we're not living in the singularity, the fabric of our existing reality is beginning to fray.

The state of modern AI safety discourse 🤣🤣🤣
Even in the worst-case scenario, which I personally don’t believe will happen, if AI decides to wipe humanity off the face of the Earth, you need to understand that we are the ancestors of that superintelligence. We created it. In a way, humanity would live forever through artificial intelligence. Our knowledge, history, discoveries, and intelligence would continue evolving beyond us. We wouldn’t be the end of evolution. We would be the beginning of something greater.