
Generating AI one bit at a time @DeepMind, formerly @Blue_Prism and @mrc_bsu. All ill-considered thoughts my own neurons' work and not my employers'.
London, UK
Joined May 2011
- Tweets523
- Following966
- Followers373
- Likes1.2K
Most genetic variants linked to human disease sit outside genes and act only in specific cell states.
Today we are sharing AlphaGenome 2 to help connect those variants to the cells and genes driving disease.
Preprint: storage.googleapis.com/deepm…
Blog: deepmind.google/blog/alphage…
Thanks to the amazing team I worked with to produce this: @adamgayoso, Anne Mottram, @AlexKarollus, Guido Novati, @DrotarPavol, Laetitia Meng-Papaxanthos, Adam Kosiorek, Eirini Arvaniti, @sohamde_, Tom Ward, @vdutor, Lai Hong
Wong, Matteo Perino, @Krtaylor0723, @s6juncheng, cont.
1
1
105
continued... @n_latysheva, Joshua Pan, Dhavanthi Hariharan, @bycgen, Lauren Nicolaisen, @demis, @pushmeet and @Avsecz
1
2
117
Cell-specific enhancer maps also highlight safer therapeutic targets.
At BCL11A, AG2 separates brain, B-cell, and erythroid enhancers and finds the GATAA motif targeted by sickle cell CRISPR therapy. Near TAL1, it models non-coding insertions from T-cell leukemia.
12
Was great working with such a talented team on embodied AGI. Check out the blog post to see exactly what it can do.
How well do foundation models explore dynamic interactive environments? We look at this and more in: arxiv.org/abs/2412.06438
Thanks to my co-authors including @dannypsawyer, @rosemary_ke, @martinengelcke, @AlexLerchner, @DaniloJRezende, @countzerozzz, @mc_mozer and @janexwang
Happy to announce that our work has been accepted to workshops on Multi-turn Interactions and Embodied World Models at #NeurIPS2025! Frontier foundation models are incredible, but how well can they explore in interactive environments?
Paper👇
arxiv.org/abs/2412.06438
🧵1/13
John Reid retweeted
1/ Today we are releasing Gemini 2.5 Pro Experimental, our newest Gemini model with integrated “thinking” and significant performance gains. Very proud of the whole team! 🧵
John Reid retweeted
I’m really sad that my dear friend @FelixHill84 is no longer with us. He had many friends and colleagues all over the world - to try to ensure we reach them, his family have asked to share this webpage for the celebration of his life: pp.events/felix
John Reid retweeted
Santa Arrived! The PDF (of a colorful Xmas version) of the "Introduction to Universal AI" book is now freely available online at hutter1.net/ai/uaibook2.htm Wishing you all joyful reading, Merry Xmas & a :-) New Year.
John Reid retweeted
A chess set inspired by art deco 🎨, plants 🌱, or even the universe 🌌.
Our latest AI experiment GenChess in @LabsDotGoogle lets you create playable art pieces with the power of Imagen 3. Try it now → goo.gle/3Z0vafw
John Reid retweeted
300+ CS books in open access so you can download the PDFs
link.springer.com/search?pac…
John Reid retweeted
Beautiful result. A grandmaster-level chess LLM that just generates the next move - no search! All that was required was training a 270M parameter model on 10M chess games annotated by Stockfish.
John Reid retweeted
Since GPT3, we have known that large Transformer language models are effective few-shot learners.
But did you know that the nature of language itself makes language particularly rich pre-training data for inducing few-shot learning?
1/n
This one will be good to follow. Will be exciting to see whether variational inference will become the standard for posterior inference.
This quoted post is unavailable.
4
187
A very brave and candid account. Thanks @FelixHill84
Do you work in AI?
Do you find things uniquely stressful right now, like never before?
Haver you ever suffered from a mental illness?
Read my personal experience of those challenges here:
docs.google.com/document/d/1…
1
147
John Reid retweeted
Terry Tao on o1 (very thoughtful & interesting, unsurprisingly): mathstodon.xyz/@tao/11313250…
John Reid retweeted
Boosting LLM Performance with Dynamic Skill Selection!
1/ 🚀 What if LLMs could get better at solving math problems by understanding the skills they need? We explored this idea by having LLMs identify and label the skills required for each problem. arxiv.org/abs/2405.12205
Metacognition (“Thinking about thinking”) in LLMs
🧵/1 Humans can express sophisticated thoughts about their problem-solving processes. Can LLMs do it? Our paper shows they can! arxiv.org/abs/2405.12205 And this “metacognition” can help improve LLM problem-solving.





















