Sabitlenmiş Tweet
Alek Dimitriev
888 posts
Alek Dimitriev
@tensor_rotator
Inference @AnthropicAI, prev Gemini @Google, prev prev PhD @UTAustin
San Francisco Katılım Mayıs 2016
1.6K Takip Edilen1.7K Takipçiler
come by Marina Green RIGHT NOW and tell us your opinion!
English
I’ve been saying it for a while, cursor’s research team is insanely high on talent density. So many people I respected from my PhD / early career ended up there. Seems like that’s bearing fruit.
English
Chuck Norris's tears cured cancer; too bad he never cried.
English
This heat wave can mean only one thing: a frontier lab is doing a huge scale up in their training run.
English
I expressed concern to @tylercowen about him traveling to South Africa due to the high crime rate.
Today, there is a bomb threat at GMU.
Tyler never misses.
English
Noam's getting involved. Let's find out how good the pentagon is at Diplomacy
Noam Brown@polynoamial
tl;dr: @OpenAI will not be deploying to the NSA or other DoW intelligence agencies for now, so that there's time to address potential surveillance loopholes through the democratic process. Over the weekend it became clear that the original language in the OpenAI / DoW agreement left legitimate questions unanswered, especially around some novel ways that AI could potentially enable legal surveillance. The language is now updated to address this, but I also strongly believe that the world should not have to rely on trust in AI labs or intelligence agencies for their safety and security. Deployment to the NSA and all other DoW intelligence agencies will be withheld so that there is time to address these loopholes through the democratic process before deployment. I know that legislation can sometimes be slow, but I'm afraid of a slippery slope where we become accustomed to circumventing the democratic process for important policy decisions. When there is bipartisan support and urgency, I have faith that government can act quickly. And as AI becomes more powerful, it's more important than ever that ultimate authority be vested in the public. I am also planning to become more personally involved with policy at OpenAI. I think now more than ever it's important for researchers to be in the loop so that policy is informed of the extremely fast progress we are seeing.
English
Alek Dimitriev retweetledi
Agree with him or not, the (oddly popular on here!) take that Dario Amodei is some kind of bumbling Silicon Valley naïf who couldn't get a deal with the Pentagon done because he doesn't understand politics seems entirely wrong.
His favorite book is "The Making of the Atomic Bomb." He used to buy copies for new Anthropic employees. (There's still a copy prominently displayed in the Anthropic library.) He fully expected -- back when it was a crazy thing to expect! -- that AI would become as important as nuclear weapons, and that the people who built it, like the scientists of the Manhattan Project, would face pressure from governments to use their technology in ways they found immoral or dangerous.
I am sure this all could have been handled differently, but none of this is a surprise to anyone who knows anything about the relevant people.
English
I have never been prouder to be working at Anthropic than today.
Anthropic@AnthropicAI
A statement on the comments from Secretary of War Pete Hegseth. anthropic.com/news/statement…
English
Alek Dimitriev retweetledi
Alek Dimitriev retweetledi
Alek Dimitriev retweetledi
i go get a heater. theres an “infrared heater”. guy says its different because it heats objects, not the air. “to heat a room” means “to heat the air”. “it heats the room, not the air”. are you buddha. “the room” is “the air”. is this the indian philosophy aisle. we are at lowes
English
@MingyuanZhou I don't disagree, I just wanted to add to the conversations that there are better and worse ways of catching up. And some of these techniques are always going to be catch up only, and never beyond that.
English
@tensor_rotator Token models with strict usage limits may not last. The challenge is rewarding quality while ensuring copycats can’t profit without fairly compensating original creators.
English
I can finally publicly state one reason I’ve not been bullish on open source catching up and overtaking the frontier labs: we observed several of the top open source models distilling from Claude. Leapfrogging happens through innovation, not distillation.
Anthropic@AnthropicAI
We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax. These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models.
English
@9K6hf1Kj6B I didn't realize it isn't innovation unless it's published
English
@tensor_rotator "Leapfrogging happens through innovation"
how many non safety papers has anthropic released? what innovation? most the innovation I see on arxiv is coming from china.
English
@tensor_rotator Then why you so worried that you want os banned?? Lmao
English
Alek Dimitriev retweetledi
