Ok, so LLMs are a Thing. How do they work? Embeddings. WTF are embeddings? I spent a year doing a deep dive. But when I was researching, I couldn't find anything that explained them in business, engineering, AND math contexts. So I wrote a thing.🚀 vickiboykis.com/what_are_embed…

📢NEW PAPER! Where Mostly here now: @davidthewid.bsky.social, Sarah Myers West & I unpack what Open Source AI even is. We find that the terms ‘open’ & ‘open source’ are often more marketing than technical descriptor, and that even the most 'open' systems don't alone democratize AI 1/ papers.ssrn.com/sol3/papers.cf…

"Episode 01010010: Regulatory Compliance Is Not The Enemy The fintech galaxy is ridden with cargo cults and myths about how regulatory standards prevent them from building software using modern best practices..." (I was going to open with this vid, but ran out of time for A/V)

For those that hope (or worry) that LLMs will do breakthrough scientific research, I've got good (or bad) news: LLMs are particularly, exceedingly, marvellously ill-suited to this task. (if you're a researcher, you'll have noticed this already) Here's why🧵

"In 1956, Massachusetts Institute of Technology (MIT) Prof Claude Shannon writes a one-page article trying to curb the information theory #hype." (terse text above by Alex Vasilescu; h/t Palle Jorgensen & Carina Curto)

arXiv -> alphaXiv Students at Stanford have built alphaXiv, an open discussion forum for arXiv papers. alphaXiv You can post questions and comments directly on top of any arXiv paper by changing arXiv to alphaXiv in any URL!

📢📢📢Vital work from AI Now Institute finding the substance behind the hype, yet again. Critical reading for anyone who's waved their hands while calling for an "FDA for AI". Here's what that means IRL, where it could work, and the ways it would likely fail w/o big changes.👇

Excited to be working on a new project this summer at Ink & Switch together with Josh Horowitz and Paul Sonnentag (he/him)! Here's a quick summary of what we're up to: # Jacquard: version control + provenance for empirical research How do researchers write papers that contain data analysis

Stunning. Incredible work. Cubic millimetre of brain mapped in spectacular detail nature.com/articles/d4158…

I've written a blog that writes this story up, and I've also created my first youtube video (!) to tell the story. I'm trying to figure out if youtube is for me so, uh, let me know. Thanks so much for reading. eieio.games/essays/the-sec… youtu.be/OI4DbECnp8A

For as along as there've been sensors, there's been noise to deal with - hence the importance of denoisers Recently we're using denoisers to solve far bigger problems -in imaging, inverse problems, & machine learning. So what are denoisers and what properties do they have? 1/n

If you think the next big idea in your field will come from inside the field, you're probably wrong.

New AI Snake Oil essay with Angelina Wang @angelinawang.bsky.social, Sayash Kapoor, and Solon Barocas — we analyze the UK's liver transplant matching algorithm that's been reported to make it nearly impossible for patients under 45 to get a liver. In this thread I'll summarize what we learned. aisnakeoil.com/p/does-the-uks…

I find this OpenTelemetry gold rush absolutely fascinating. Even if maybe most of those newfangled solutions will fail, the collective investment will be really interesting about what it does to programming ecosystems. It will just have to make async locals and co. more popular.

Posting a call for help: does anyone know of a good way to simultaneously treat both POTS and Ménière’s disease? Please contact me if you’re either a clinician with experience doing this or a patient who has found a good solution. Context in thread

I'm giving a talk about One Million Checkboxes, scaling sites with simple tech, and the philosophy that I bring to stranger-interaction projects! It'll be at the Recurse Center in late January. Come hang out~

Today, we are publishing the first-ever International AI Safety Report, backed by 30 countries and the OECD, UN, and EU. It summarises the state of the science on AI capabilities and risks, and how to mitigate those risks. 🧵 Link to full Report: assets.publishing.service.gov.uk/media/679a0c48… 1/16

One Million Applause for One Million Checkboxes 👏✅👏✅👏 Check out nolen's brilliant talk on games, connecting strangers, and how some very smart teens subverted his code and touched his heart ❤️‍🔥

We've seen survival times for various cancers triple in the last two decades due to advancements in monoclonal antibodies and CAR-T cells. All those ideas, proofs of concept, and initial clinical trials came from academia.