We first saw evidence for this ~ 2019. We were super surprised & skeptical. After 5 years of quintuple cross-validation, we feel confident to share with the world.

Absolutely delicious community note.

a distinct disipline is emerging from the bio & ML intersection -- predictive biology - enabled by AI - understanding = prediction, not causal diagrams - emergent instead of reductionist - new institutions: industrial > academic a rough attempt at outlining the new field:

Literally the single biggest reason we built cephla.com

The UK needs more science and innovation. Excited to be co-hosting this event with ARIA & Wilbe.com on Sep 10 evening at Wilbe’s London HQ. Join Devika Thapar , Ilan Gur, Thomas Kalil, and a community of innovators to explore wild ideas. lu.ma/4s0aalf0

This is the coolest "application" of LLMs I have seen in a while: instead of using an LLM to solve a problem set, flip the table and ask the student to TEACH the LLM how to solve the problem set. Really lovely. Thx to Jordan Hoffmann for sharing this :-). youtube.com/watch?v=om7VpI…

Very cool to see neuro-endovascular approaches expand volumetric access to the brain. + recent work from Medtronic assessing the feasibility of endovascular DBS in the thalamus (link below). Still lots of work to do but exciting times for Interventional Neurotechnology.

Looking forward to this: Tom is a legend and this will be a fun conversation

LLMs that incrementally synthesize formally verifiable reimplementations of legacy code will not just solve cybersecurity, they will also be good for performance. Electron apps can be gradually converted to WASM 🦀

NIH is asking you to think about what makes data sharing useful . Wonderful.

My view on this has not changed in the past eight years: I have given many talks and written position paper in 2019 (link below). Progress is faster than my past expectation. My target date used to be ~2029 back then. Now it is 2026 for a superhuman AI mathematician. While a

The UK’s new moonshot research agency just launched an £81 million program to develop early warning systems to sound the alarm if Earth nears climate tipping points for the Greenland Ice Sheet and a massive gyre in the North Atlantic. technologyreview.com/2024/09/04/110…

As we move towards more powerful AI, it becomes urgent to better understand the risks in a mathematically rigorous and quantifiable way and use that knowledge to mitigate them. More in my latest blog entry where I describe our recent paper on that topic. yoshuabengio.org/2024/08/29/bou…

The new Dynamicland website launched today. Our vision for transforming the science lab with communal computing can be found on the right of the third shelf. More to come.

getting a bag of Doritos…

Very cool to see Deepmind protein design work use UK's Diamond Light Source to experimentally validate their structures IMO Diamond is a jewel in the crown for UK science capacity that should be celebrated much more Not many AI and sci policy ppl are aware of this facility (1/3) 🧵

Can we all have a moment of sympathy for DNA – it has had a rough week 🧬 First, proteins not predicted by genetic code 😮 biorxiv.org/content/10.110… Then, inheritance without genes through amyloid protein 🤯 nature.com/articles/s4155… Biology is amazing!

"The apparent tuning of hippocampal neuronal spiking to different sensory modalities might emerge due to alignment to the afforded action progression within a task rather than representation of external cues." Most rat studies = foraging Alignment is to room, thus place cells

Correct: ee.stanford.edu/news/2021/jan/…

I'm deeeeeep into water & cement at the moment. Water is the #1 most used product in the world. Cement is the #2 most used product in the world. The scale is mind-blowing. Scale is on another level when you're in these two markets.