i don’t “taste” ice cream flavors, i just pick one and commit to it

i am still of the opinion that solving the boring operational problems in biopharma r&d, like slowness in regulatory documentation and process development is the most effective way to derive value from AI in the short term

this 1785 paper by william withering is probably one of my favorite medical documents for its extreme intellectual honesty. it is also probably the first systematic documentation of dose-response relationships in medicine. i'll quote one of my favorite paragraphs here but the

a highly underrated book. one of my favorite reads this year a.co/d/9IaOxdN

Baby KJ 😍😍

This summer, the world learned that an infant -- baby KJ -- had been saved by gene editing. This morning at #STATSummit, we had KJ and his family on-stage. If you can watch this clip without tearing up... you may need a gene related to empathy edited. We're having an amazing

please stop seeking AI’s advice on interpersonal relationships. think and feel for yourself

this book is very good

if you are Nabokov pilled, read Ada!!!

A lot of you have been DMing me asking about my peptide stack, so here it is: 10% luck 20% skill 15% concentrated power of will 5% pleasure 50% pain 💯 reason to remember the name

Excited to release BoltzGen which brings SOTA folding performance to binder design! The best part of this project has been collaborating with many leading biologists who tested BoltzGen at an unprecedented scale, showing success on many novel targets and pushing its limits! 🧵..

everybody wants to build AI scientists but nobody wants to do the hard work of standardizing ontologies to make this happen

What do Starlink, the trad wife, and Marshall Mcluhan have in common? Introducing the Third Oikos, a new series about how technology is reshaping the household. thethirdoikos.substack.com

LLMs memorize a lot of training data, but memorization is poorly understood. Where does it live inside models? How is it stored? How much is it involved in different tasks? Jack Merullo & Srihita Vatsavaya's new paper examines all of these questions using loss curvature! (1/7)

wait the biosphere would be mostly fine if we got rid of all the mosquitoes?

You have heard of AI slop in the context of short video creation. But the same principle applies when it comes to improving drug discovery: we absolutely do not need a deluge of new hypotheses; we need better predictive validity (as per Jack Scannell). writingruxandrabio.com/p/what-will-it…

Starting soon at 8:45AM! All the talks for AI+Science conference at Caltech will be livestreamed and also available later. aiscienceconference.caltech.edu

I’ll be at NeurIPS Dec 2-4 and at the Foresight Institute event in SF Dec 5-7. If you’re working on AI in bio— may I meet you?