Google DeepMind breaks up the AlphaFold team as research focus shifts

In short: Google DeepMind has reassigned and lost many of the researchers who built AlphaFold, as it shifts toward broader AI tools for science.

What's going on

Google DeepMind has dismantled the team behind AlphaFold, its AI system that predicts the 3D shape of proteins. Proteins are tiny building blocks in the body, and their shape matters because it affects how they work (like how a key’s shape decides which lock it fits).

An analysis by the Financial Times found most original AlphaFold paper authors have been moved to other roles over the past year. DeepMind confirmed many are now working on projects built around Google’s Gemini large language model, plus areas like enzyme design, nuclear fusion, and genomics.

Some researchers moved to Isomorphic Labs, Alphabet’s drug discovery company that was created in 2021 to turn AlphaFold’s progress into real-world medicine work. Nearly a quarter of full-time AlphaFold paper authors have left DeepMind entirely. John Jumper, a key AlphaFold leader, and other researchers also moved to internal efforts to improve Google’s AI coding, and Jumper later said he is leaving for Anthropic.

What to watch

DeepMind says it is moving from teams focused on one big scientific problem to building Gemini-powered systems that can help scientists with many tasks, and possibly automate parts of research over time. The next signal to watch is whether this leads to new widely used science tools, or whether the loss of well known AlphaFold talent slows DeepMind’s pace as competition for top AI researchers heats up.

Source: Financial Times