🔥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

It will be hard to keep the commanding heights in (western) liberal hands while simultaneously handing the technology over to the governments of the developing world. These goals are probably not compatible—it will take a lot of work to make them so.

would be great if we had a way to robustly make AI do (only) what we want it to do

Introducing an upgraded Claude 3.5 Sonnet, and a new model, Claude 3.5 Haiku. We’re also introducing a new capability in beta: computer use. Developers can now direct Claude to use computers the way people do—by looking at a screen, moving a cursor, clicking, and typing text.

New paper! How should we make trade-offs between the quantity and quality of labels used for eliciting knowledge from capable AI systems?

Someone recommended The Goal from Goldratt to me and I have to say, there's nothing in there that Factorio hasn't taught me already

i've played that game in my dreams before

Video summary for "Prover-Verifier Games improve legibility of LLM outputs" youtu.be/EMDa4urzz-M 1/2

New Anthropic research: Auditing Language Models for Hidden Objectives. We deliberately trained a model with a hidden misaligned objective and put researchers to the test: Could they figure out the objective without being told?

New paper with Johannes Treutlein , Evan Hubinger , and many other coauthors! We train a model with a hidden misaligned objective and use it to run an auditing game: Can other teams of researchers uncover the model’s objective? x.com/AnthropicAI/st…

i enjoy being a hypocrite as much as the next person, but if there’s one thing i can’t stand it’s a hypocrite

Introducing the next generation: Claude Opus 4 and Claude Sonnet 4. Claude Opus 4 is our most powerful model yet, and the world’s best coding model. Claude Sonnet 4 is a significant upgrade from its predecessor, delivering superior coding and reasoning.

Oh my god. I’m so fucking relieved and happy in this moment

Do not go gentle into that good night

There is still an opportunity for OpenAI to live up to its founding promises, instead of abandoning them. Here I explain what this could look like.