igelblork retweeted
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com/
857
2,595
19,577
4,297,101
igelblork retweeted
I will not “lock in”. I will not “looksmaxx”. I will not “ascend”. I will not “cold approach”. I will wait for a woman to initiate something. And if that doesn't happen? I kill myself.
70
1,263
17,336
210,395
haven't seen @avgvstvsczr around.........
1
91
it's very difficult to pick 9 so i resolved to pick "9 that i loved during my formative years"
2
1
4
216
David Sacks compares banning AI to the Chinese emperor who banned shipbuilding
“If you go back to the medieval time period, Chinese civilization was much more advanced than European civilization. And somewhere along the line, it basically switched and the West basically took the lead. And some historians have pinpointed this to the decision of a single Chinese emperor to ban shipbuilding."
“What that allowed is the Europeans then basically colonized the whole world and all the riches that flowed from that flowed back to Europe. And that's how Europe took the lead. And I think that if we were to do the Bernie Sanders thing, which is basically ban AI, it's like banning shipbuilding."
“Chinese civilization will rocket past us, they will be the ones that make all the discoveries and discover this new world and accrue all the wealth. But that's what we're talking about doing. I mean, crazy things like this have happened in history, where you'll engage in self sabotage."
57
268
1,523
95,303
igelblork retweeted
You probably know nothing about AI development. That's okay. You can discern everything you need to know by AI's enemies: Bernie Sanders, Epstein's best friend (Bill Gates), and the man who bankrupted California.
16
24
207
4,763
20 irrelevant countries with no tech industry (among them mine). i wonder if the leadership in these countries have any level of regret for the self-sabotage that led to this moment.
SITUATION DETECTED: World leaders from Canada, Australia, Germany, the EU, the UAE, and others have signed a letter calling for AI control mechanisms based on Dario Amodei’s essay, per Politico.
1
130
okay "sally stockholm"
Getting a like from the co-founder of OpenAI means the world to me.
Thank you @gdb
I pour my entire life into my research and making it available to you all. ❤️
1
141