🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

🤖 !! Model-free agents can internally plan !! 🤖 In our ICLR 2025 paper, we interpret a model-free RL agent and show that it internally performs a form of planning resembling bidirectional search.

AI-2027 is live! Finally. What's next? --Making bets with people who disagree with us --Awarding prizes to people who write alternative scenarios --Awarding prizes to people who convince us we were wrong or find bugs x.com/DKokotajlo/sta…

Is there precedent for this kind of thing? It’s a total bait and switch. If this were a paper it would be desk rejected.

Well, but what if it didn’t? If our options are: - extinction - totalitarianism - stop AI progress until we get our shit together Doesn’t the 3rd option seem best?

How do you identify training data responsible for an image generated by your diffusion model? How could you quantify how much copyrighted works influenced the image? In our ICLR oral paper we propose how to approach such questions scalably with influence functions.

It's crazy that we don't know why Sydney Bing was so unhinged. While I bet there are people like j⧉nus who have a good behavioral understanding of Sydney, as far as I know we have no mechanistic understanding of those behaviors (and most other AI behaviors)

How might your life "have to" change if you thought AI x-risk was higher? or lower?

It is sad to see OpenAI's mission being reinterpreted to mean "proliferate OpenAI's products among non-profits". This is not the mission articulated in the OpenAI Charter, which it championed for years internally. It is the least onerous alternative that still says "non-profit".

A potentially useful analogy for AI doom is COVID -- I was ignoring it for a while, even though I'd heard about it. It took a personal message from someone I trusted to make me concerned enough to want to look into it more.

Why didn't the (real) sharing economy take off more? Like, why don't more people rent out their houses/cars/computers/whatever when they're not using them? I guess it's mainly just that it's a bit of a hassle to set-up and kind of stressful to deal with / manage.

By the time we have clear and obvious evidence that AI poses an existential threat to humanity, it will almost certainly be too late.

Today, myself and 11 other former OpenAI employees filed an amicus brief in the Musk v Altman case. We worked at OpenAI; we know the promises it was founded on and we’re worried that in the conversion those promises will be broken. The nonprofit needs to retain control of the

I am happy to announce that I have joined the White House Office of Science and Technology Policy as a Senior Policy Advisor on AI and Emerging Technology. It is a thrill and honor to serve my country in this role and work alongside the tremendous team Director Michael Kratsios has built.

🚨BREAKING🚨 OpenAI's top official for catastrophic risk, Joaquin Quiñonero Candela, quietly stepped down weeks ago — the latest major shakeup in the company's safety leadership. I dug into what happened and what it means for Obsolete 🧵

I've got some big personal news: I'm joining Goodfire to lead a fundamental interpretability research team in London! This has been a while coming /n

The argument for worrying about AI takeover is strikingly similar to the argument for worrying about AI-enabled coups. Yet there is MUCH less work done to prevent AI-enabled coups. That needs to change!

The way AI companies talk about "commitments" is downright Orwellian. They aren't actually committing to anything, and they change them all the time.

🧵 Today we publish a comprehensive report on "AI Behind Closed Doors: a Primer on The Governance of Internal Deployment". Our report examines a critical blind spot in current governance frameworks: internal deployment.

Disappointed, not surprised tho

Very cool to see gradual disempowerment in 80k's problem list! 80000hours.org/problem-profil…