Podcast Episodes

“You Should Apply to Inkhaven” by Tomás B.

Inkhaven is a writers residency in Berkeley, in which the only requirement is you have to publish 500 words each and every day. Though I always had …

8 hours ago

“Pretraining data, not verifiability, is why LLMs are especially good at math (and coding)” by Steven Byrnes

Follow-up to: “LLMs are (still) mostly powered by imitative learning, not RL”

A common take I’ve been hearing is: “LLMs are especially good at math …

18 hours ago

“You don’t need a union to go on strike” by sudo-nym

I'm mostly hoping this somehow gets sent to a privately disgruntled frontier lab employee, but it would also be cool to expand other people's minds …

18 hours ago

“Stopgap Measures to Address Immediate AI Security Threats” by Andrea_Miotti, Gabriel Alfour

Most people know AI as the technology behind chatbots like ChatGPT. However, what the top AI companies are explicitly aiming for is something else e…

18 hours ago

“The Preference Cascade Is Only Getting Started” by Zvi

We are in the midst of a preference cascade about existential risk from AI.

A preference cascade is, alas, the best method we have to change the de…

20 hours ago

“The Horse” by Character#2736

You have a horse.

You do not like the horse. The horse does not like you.

At the moment, you are completely dependent on the horse. The terrain is …

21 hours ago

“The Game is Set for a Targeted Memetic Attack on the AI Safety Community” by keltan

While this is relevant to my work at MIRI, I have not checked these ideas with anyone else on the team and am posting this on my personal LW account…

1 day, 4 hours ago

[Linkpost] “Three Hackers used Opus 5 to Hack Into OpenAI’s Core Codebase [WSJ]” by Linch

This is a link post.

Three whitehack hackers from Hacktron used Claude Opus 5 within hours of release to chain exploits into hacking to OpenAI's mono…

1 day, 6 hours ago

“Deep recurrent models are less robustly CoT-monitorable than normal CoT models in a toy setting” by Nick Kuhn, Alek Westover

We use RL to teach a deep recurrent model and a normal CoT model to solve a math problem while hiding from a CoT monitor which of two possible probl…

1 day, 8 hours ago

“Superintelligence this Christmas” by Alexander Gietelink Oldenziel

I now consider it plausible that some form of recursive self-improvement is imminent, and that we may be on track for superintelligence by Christmas…

1 day, 10 hours ago