
Tim Rocktäschel
5.9K posts

Tim Rocktäschel
@_rockt
Co-Founder @Recursive_SI, Professor of AI @AI_UCL, PI @UCL_DARK, Fellow @ELLISforEurope. Ex @GoogleDeepMind @AIatMeta @CompSciOxford


Project Genie is magical... but we've also been working on some new ways to interact with another player (or agent). It was super fun to demo this new capability at Google I/O this week, where we enabled attendees to explore worlds with Gemini as a companion. Going forward, we are incredibly excited to see how this can enable Gemini to learn how to interact with humans in embodied environments. Some examples of interacting with Gemini in real-time within these generated worlds:






Real-world models are here! Stoked to share how we're bringing real-world locations to life by integrating Street View into Genie. Try it now at labs.google/fx/projectgenie and read the blog for more info: blog.google/innovation-and…


Inference Scaling We think our results are just a lowerbound on whats possible today. As @polynoamial puts it, benchmark numbers are a function of inference compute. Tokens🤝FutureSim📈 We think there's potential to design better harnesses beyond coding that push the agents further.

It was such a privilege to join cofounders, @_rockt and Alexey Dosovitskiy, and a phenomenal team to celebrate @Recursive_SI’s launch and $650m raise last night. I have been struck by Tim’s exceptional ambition, experience and humility in the last few months. I am delighted he and team will be pushing the frontier of AI right here in the UK 🇬🇧

Sometimes I miss the days when people were passionately fighting about MCMC versus variational methods, or whether posterior tempering is problematic. We should have a nostalgia ICML 2010. You can submit AI slop, but expect a 2010 era reaction. What happened to our field?



For the past few years, we’ve been thinking about the impact of open endedness on the pursuit of AGI. Through our exploration, we’ve met some of the most talented researchers and engineers automating scientific discovery. When all of our “top people” decided to start a company together, we knew we had to be involved. Coding agents may help implement experiments but the step to building automated research agents is going to be hard; harder than most anticipate. This team can figure it out. And…I always back Josh. You’re stuck with me :)


Excited to co-found Recursive (@recursive_si) with an exceptional team in London and SF to create AI that experiments on how to safely improve itself, turning compute into knowledge that accumulates in an open-ended process of endless, automated scientific discoveries.









