Today's "DeepSeek selloff" in the stock market -- attributed to DeepSeek V3/R1 disrupting the tech ecosystem -- is another sign that the application layer is a great place to be. The foundation model layer being hyper-competitive is great for people building applications.

A brain-reading implant that translates neural signals into audible speech has allowed a woman with paralysis to hear what she intends to say nearly instantly go.nature.com/4iNk8l0

Scientists have developed a device that can translate thoughts about speech into spoken words in real time. Although it’s still experimental, they hope the brain-computer interface could someday help give voice to those unable to speak.

Work led by BAIR students Kaylo Littlejohn and Cheol Jun Cho advised by BAIR faculty Gopala Anumanchipalli "...made it possible to synthesize brain signals into speech in close to real-time." dailycal.org/news/campus/re… via The Daily Californian

Humans just saw a *new* color—literally outside the known visual spectrum. BAIR faculty and visual computing expert Ren Ng and collaborators made it possible with the Oz Vision System. 🌈👁️ Newly published in Science Advances: science.org/doi/10.1126/sc… popsci.com/health/new-col…

Excited to build on pioneering work from Frank Willett, Sergey Stavisky, and Eddie Chang labs showing 3–4x faster communication (60–80 words per minute) to approach natural speech speed!

being good at ML systems helps you run more experiments. being good at ML theory helps you run less experiments

This is a truly exceptional result. An instantaneous brain-to-voice synthesizer with paralinguistic features like prosody. Congratulations Maitreyee Wairagkar BrainGate Team on the monumental achievement!

I’m excited to share our new nature paper 📝, which provides strong evidence that the walkability of our built environment matters a great deal to our physical activity and health. Details in thread.🧵 nature.com/articles/s4158…

Excited to share our new study on inner speech out now in Cell Press! W/ Benyamin MK (co-lead) Frank Willett (senior) doi.org/10.1016/j.cell…

my biggest codex power user tip is to use dictation install macwhisper or whatever you like you'll just end up specifying like 3× more than you would by typing and avoid so many footguns just try it

📢 Please retweet: We're recruiting **PhD students and Postdocs** at UC Berkeley & UCSF! Please apply if you are interests in AI for healthcare, statistics/causal inference, or medical vision-language models. For more details, check out this link: forms.gle/1rZHBoH8f8fu27…

inference time learning is imo a more accurate term than continual learning, otherwise continual learning is just finetuning