Cat retweeted
Two weeks from delivery (and 5 billion tokens!) to getting this engine online. I could not be more proud of my team.
It’s so fucking cool.
Happy friday everyone!
Cat retweeted
Just called out one of the most egregious cases of stolen valor I’ve ever seen.
Uniform all fucked up, hair WAY out of regulation, and completely the wrong age range.
And the NERVE to ask me for candy?
Get the fuck out of here, child. No way you ever served in the 82nd Airborne.
Cat retweeted
We have signed an agreement to acquire Rune Technologies, bringing purpose-built super intelligence to logistics at the tactical edge
Together, we'll connect frontline operations with enterprise workflows to strengthen mission readiness, even with limited connectivity
More: sforce.co/4i92V7W
Cat retweeted
Introducing DEEP AMERICA: A Subterranean Plan for the Nation. Strider Capital presents Phase 3 of the National Security Renaissance plan.
This is a proposal to build America’s defense, civil, and youth institutions underground. /1
Cat retweeted
Introducing Kino: make videos end-to-end with Claude Code and Codex.
Drop your footage, get highlights, edit together, send links for feedback, and let your agents cook.
AI is becoming a great creative partner. Kino is a new home for your team and agents to collaborate.
Cat retweeted
i took @TheRealDanSaedi to the park to talk about some AI stuff and why the markets keep going down.
we covered:
- the bear case for AI
- leverage
- why the app layer is good actually
00:00 - Guessing Daniel’s race
01:01 - If you’re not close to the singularity you’re ngmi
02:11 - Congrats @rmcentush and @CatOrman1
03:30 - Advantages of building AI in NYC
05:19 - AI probably won’t kill you
11:20 - The case against the AI exponential
15:32 - We are committed to this path (scaling laws via transformers leads to super-intelligence)
16:58 - The window for the bubble popping is opening
22:58 - As the frontier goes up the tasks stay the same
26:08 - Leverage and high rates are a risk to AI
30:39 - It’s going to be Accenture that puts AI into regular people’s hands
Cat 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/
feeling extremely grateful & surrounded by the people we love
@rmcentush and I had a great weekend :)
🤖 Made with AI
