@catehall there are three ways to get exposure:
- get hired
- marry someone who was hired
- be a connected VC/LP
honestly not sure which of the 3 is easiest
English
I take you seriously
6K posts
I take you seriously
@replyallguy
I feel self conscious conversing on X because it pollutes the feeds of my followers. Hence this burner account. Only follow if you enjoy my burner’s pollution.
San Francisco, CA Присоединился Şubat 2021
586 Подписки157 Подписчики
@CharlesD353 @occamsbulldog obviously they’re doing inference work to bring the cost down and will release later this year. cost and capacity are not constants.
English
@occamsbulldog Since they decided to train it, in September(?) they've seen a big positive trend break in demand and have to triage compute more than expected.
I agree it's not the most likely single explanation, but it probably made the decision easier at least.
English
The Boys (TV show) would be great if it was like 80% less gross
English
English
As I've talked to folks about Thinky's purpose and why Thinky exists, I've found myself referring to The Intelligence Curse (intelligence-curse.ai) time and time again. So much so that I've even suggested in the past that The Intelligence Curse would serve pretty well as Thinky's mission statement :)
It's almost uncanny how much @luke_drago_ and @LRudL_ 's mission on keeping humans empowered matches @thinkymachines' mission. Excited to work together!
Workshop Labs@WorkshopLabs
Workshop Labs is joining @thinkymachines. We believe there's a path for AI to make humans matter more. We couldn’t be prouder to join Thinking Machines to see this work through. workshoplabs.ai/blog/wsl-joini…
English
@michael_nielsen @stevenstrogatz grandmasters rotate opening a lot, and fashions change year by year. i wonder what the control looks like.
English
This problem of extracting explanations from large models is interesting. Magnus Carlsen's team seems to have benefited (a lot) from some phenomenology of AlphaZero
Dwarkesh Patel@dwarkesh_sp
A deeper question about AlphaFold: In what sense is even a scientific explanation at all? It’s not going to have some crazy unforeseen predictive power outside of protein folding, such as say, relativity had on Mercury’s precession. There's a couple of ways you can interpret this: 1. it's not an explanation at all, 2. it contains little explanations you can extract through interpretability (Magnus Carlsen appears to have changed his game after AlphaZero forensics were published), or 3. it's a genuinely new type of scientific object we don't have the verbs for yet.
English
@MaxCRoser @robertwiblin don’t they plan to release it once they get inference costs down?
English
Anthropic's new model, Claude Mythos, is so powerful that they're not releasing it.
They must have good reasons. They're willing to forgo a lot of revenue.
@robertwiblin on what we know so far about how dangerous the current situation is.
80000hours.org/2026/04/claude…
English
@mooncat_is @tensor_rotator yeah seems like we’d need the false positive rate from hinting that every sniper has a bug
English
“We took the needle the model found, isolated the relevant handful of the haystack, and then gave it to a small child, who found the needle as well.”
clem 🤗@ClementDelangue
"But here is what we found when we tested: We took the specific vulnerabilities Anthropic showcases in their announcement, isolated the relevant code, and ran them through small, cheap, open-weights models. Those models recovered much of the same analysis. Eight out of eight models detected Mythos's flagship FreeBSD exploit, including one with only 3.6 billion active parameters costing $0.11 per million tokens. A 5.1B-active open model recovered the core chain of the 27-year-old OpenBSD bug." aisle.com/blog/ai-cybers…
English
@MatthewJBar @KelseyTuoc I suspect saying you forgot everything is supposed to be humble and signal that you’re not threatening. When you actually meet someone who doesn’t know the capital of the USA, it’s disappointing and you wish that they remembered more from school.
English
@KelseyTuoc It's pretty hard to motivate kids to deeply learn subjects that they personally find very boring. I'm totally in favor of giving kids the opportunity to learn whatever they want, but I'm highly skeptical that schooling has much value beyond its signaling and daycare functions.
English
There's a strange disconnect in how people talk about school. Most people insist it's very important for kids to stay focused in school and learn the material, yet most adults admit to having forgotten nearly everything they learned in school beyond the basics.
English
@Miles_Brundage i’m a fan of the wait and see approach to AI regulation
English
Me when someone implies that we can just wait til next year to pass federal AI legislation
English
@Miles_Brundage real, absolutely. important? hard to say.
English
People who know things, is this Claude Code leak thing real / important?
English
@Kasparov63 humans are actually way less than 100%. 100% score is impossible even for superintelligence as getting max scores on some levels requires lucky guessing.
English
@peterwildeford @LeadingFutureAI @AlexBores first was paywalled so unfortunately couldn’t read it. second didn’t seem to support the assertion. i’m very open to the possibility, but still open to the possibility it’s not true. for the record, i’m not a fan either but still seems plausible it’s a brockman thing not openai.
English
English
When the OpenAI - Andreessen - Palantir Super PAC @LeadingFutureAI first went after @AlexBores, Bores was only 10% on the markets. They were picking on a major underdog.
Now @AlexBores is at 35% to win. Seems like the plan is really backfiring!
English
@peterwildeford @LeadingFutureAI @AlexBores where’s the source for Lehane advising Brockman? was it an individual donation or company donation?
English
@replyallguy @LeadingFutureAI @AlexBores Accurate-ish. The $25M donation came from OpenAI President Greg Brockman as advised by OpenAI Chief Global Affairs Officer Chris Lehane, and the campaign playbook is the same as Chris Lehane's previous playbooks.
English
@baym @eastouest OpenAI fund is north of $100B. If they ramp to giving 5% a year, that’s roughly NSF R&D spend (7 bil last I checked)
English
This must sound so awesome if you don’t understand the difference between million and billion
roon@tszzl
the private sector has been remaking its own versions of NIH, ARPA etc as these public science institutions have seen structural decline and defunding and it will be supercharged by the funding NPV of machine intelligence and its firepower at allocation decisions
English
@davidmanheim i disagree but i see your pov. thanks for engaging ♥️
English
@replyallguy @OpenAI 2. You're kidding yourself if you don't think OpenAI could find buyers at their current valuation. They'd prefer not to, and keep more equity to boost later valuation but that's literally violating their nonprofit mission by doing less to prioritize protecting the for-profit!
English
The @OpenAI Foundation promised State AGs and the public that in exchange for becoming a for-profit, they would spend 27% of the company's value charitably.
They need to actually do that.
Here's a post explaining why they need to go much bigger, quickly.
English
@TheZvi “we won’t push the frontier because safety is so important” 💀
English
So, Anthropic peoples, it's a great idea, but why is this in practice a safe and non-insane thing to be doing? What is your protocol for rendering it safe and non-insane?
Claude@claudeai
You can now enable Claude to use your computer to complete tasks. It opens your apps, navigates your browser, fills in spreadsheets—anything you'd do sitting at your desk. Research preview in Claude Cowork and Claude Code, macOS only.
English
Life update: after 7.5 years, I’m leaving @coeff_giving. I love it, always have. When I joined we were small, and last year we gave away over $1 billion to charity.
I helped fund science alongside some of the most thoughtful, brilliant people I know. Time to pass the torch ❤️
English