Keep on to blur preview images; turn off to show them clearly
how have people who post about proofs of novel mathematical results from large language models not yet learned to wait like five min for a proper expert to take a look before they yet again embarrass themselves
gpu enjoyer at @modal. he/him. ex @full_stack_dl, @weights_biases (acq. @CoreWeave), phd Berkeley @Redwood_Neuro. try https://t.co/SYWVMCazZ3
Black Friday bonuses, and hints to November squares available through Cyber Monday AoE, QED!
Market Design/Entrepreneurship Professor @HarvardHBS & Faculty Affiliate @Harvard Economics; Research @a16zcrypto; Editor @restatjournal; Econ @Quora; … | #QED
Three years ago, @1337skulls launched what I didn't know at the time would be both my closest and longest-running NFT community, and the start of my own artistic explorations. Today's "45 squared" edition is a tribute to 1337: the pulse of Mega City. And in the tradition of 1337 Skulls, this token is CC0. (And of course, there's a puzzle.) "1337 15 3" will be gifted to active 532s (i.e., the collective that runs 1337 Skulls), which means that ~25 are available for private sale at the thematic price of .04545; please DM if interested. And as always, a few will be reserved for puzzle solvers. @flashrekt, @51991eth, @Maniinng, @SteveRunsEth, et al.: Thanks for being 1337 – you're an 1n5p124710n, Q3D 🙏💀🔲 PS: Black Friday sale still in effect, details below 👇👀
Black Friday bonuses, and hints to November squares available through Cyber Monday AoE, QED!
RT @StefanFSchubert: Today it's three years since ChatGPT was launched. Those claims are definitely falsified by now.
We're in a race. It's not USA vs China but humans and AGIs vs ape power centralization. @deepseek_ai stan #1, 2023–Deep Time «C’est la guerre.» ®1
Seems like the Erdos problems are both an unlimited source of fun and unlimited source of drama 🤣. Turns out the problem that Harmonic solved yesterday was the "easy" version of #124 and the hard version remains open! (Funnily by the way the lean formalized version --formalized by humans-- was even weaker than the "easy" version, but that's a story for another time). As far as Erdos-AI problems are concerned, #848 that we describe in the gpt-5 paper remains the cleanest, solved by a collaboration AI-humans. And as far as >10 years old open problem solved fully with AI I think Chapter IV Section 3 is a good example, it's only open since 2012 but about O(10) humans looked at it, each for several weeks. And crucial to always remind: this is only useful insofar as it is useful to actual human mathematicians. There is no other point than that.
I work on AI at OpenAI. Former VP AI and Distinguished Scientist at Microsoft.
If you’re building complex software, this is a great way to allow agents more insight. I did this for Podscan, too. I recorded myself going through the app, page by page, every component, and had Claude turn the transcript into a knowledge document. Loads with each prompt.
Building https://t.co/od97B0HVrk and https://t.co/666FnyVVE0 in Public. Raising all the boats with kindness. 🎙️ https://t.co/6w69DZmi8H · ✍️ https://t.co/lpnor5rsTW