literally no one asked for this technology

OpenAI has already caught up with the latest feature in Claude, and also has an advanced Golden Gate Bridge mode based on sophisticated mechanistic interpretability research.

Woah:

“The page had three signatures. All three of them were Sam Altman.“

her

By the way, while some of the legal stuff in the OpenAI NDA situation is very complicated, this particular bit of nonsense is unambiguously not the way the law works: vox.com/future-perfect…

I’m once again reminded that if “alignment” really is a thing, then we probably need to solve it for capitalism before we have a chance of solving it for AI.

T H A N K Y O U ! ! !
(v 2.0)

Full RAG in 20 lines.

This is how you implement semantic search in ~10 lines

Replace 𝚌𝚘𝚗𝚝𝚎𝚡𝚝𝚜 & 𝚚𝚞𝚎𝚜𝚝𝚒𝚘𝚗 with your own.

* the blurred code is for loading example contexts

(Everything from FastAI's guide:fastai.github.io/lm-hackers/lm-…)

I can’t begin to tell you how many founders hear the words “retrieval-augmented-generation” and think they need to hire PhDs

It’s kind of impressive that an entire industry is pulling of this heist

RAG is a misunderstood term: it’s not a unified system, it’s Retrieval then Generation.

“My RAG isn’t working” is a vibe, user-felt thing.
It’s akin to “my car isn’t working”: this is how the user sees it, but you’d be pretty annoyed if that was what your mechanic said to you.

impressive follow-up from Kelsey Piper bringing receipts that it wasn't just some standard exit paperwork from OpenAI;
they seem to have proactively pressured employees to sign the non-disparagement clause, with crystal-clear threats to exclude them from future liquidity events

mmm'kay...

Scoop: OpenAI's senior leadership says they were unaware ex-employees who didn't sign departure docs were threatened with losing their vested equity. But their signatures on relevant documents (which Vox is now releasing) raise questions about whether they could have missed it.

When JJ makes software, you should pay attention. He's moved beyond focusing on R, and is building for the broader ecosystem now

He recently led the creation of Quarto which is an absolute banger. Now he's going to introduce his new LLM eval tools!

Prof Deborah Lupton Master of Public Health, PhD David Fisman Amanda Kvalsvig Raina MacIntyre 😷💉💉💉💉 Shovon Bhattacharjee, PhD Joe Kirby Institute, UNSW Mark Ungrin Matthew Oliver And here's our piece for The Conversation - also just out (you can syndicate): 48/
theconversation.com/masks-work-our…

It’s out! Our new state-of-the-science review of MASKS/RESPIRATORS in reducing transmission of respiratory infections. 13 authors (for our disciplines, see posts 3-4). 38000 words. 413 references. One conclusion: these devices work. For detail, read on. 1/
journals.asm.org/doi/10.1128/cm…

A new gold standard review of 100 studies finds:

- masks are effective in reducing transmission of respiratory diseases

- N95 respirators are significantly more effective than medical or cloth masks

- mask mandates are effective in reducing community transmission

OMG, it works so much better than I ever could have expected.

I love AI

Yes! My website poisoning attack works on Google's new LLM-powered AI overviews!

I gave my notice to OpenAI on May 14th. I admire and adore my teammates, feel the stakes of the work I am stepping away from, and my manager Miles Brundage has given me mentorship and opportunities of a lifetime here. This was not an easy decision to make.