The Financial Times (FT) reported on the 29th that Google DeepMind disbanded the team that developed AlphaFold, the protein-structure-prediction artificial intelligence (AI) model that produced a Nobel Prize winner. With large language models (LLMs) and AI agents on the rise, the research organization and strategy were reorganized around Gemini.
According to FT, most of the co-authors of the AlphaFold paper were reassigned to other departments over the past year. DeepMind said they moved to research areas including Gemini-based projects, enzyme design, nuclear fusion, and genomics.
Some researchers moved to Isomorphic Labs, Alphabet's drug discovery subsidiary. One-quarter of DeepMind's full-time authors who worked on the AlphaFold paper have left the company.
FT analyzed that the reorganization shows a significant shift in DeepMind's research strategy since the emergence of LLMs. Whereas in the past the company concentrated research staff on solving single, concrete scientific challenges such as protein folding, it now plans to focus on developing general-purpose AI systems that can be applied across multiple scientific fields.
Pushmeet Kohli, vice president of research who leads the "AI for Science" team at DeepMind, said, "For the past nine years, our strategy has been to focus on grand challenges—specific goals that every project could rally around," adding, "Now that strategy has evolved."
DeepMind plans to focus on building Gemini-based systems that can support scientists' research and, in the long term, automate parts of the scientific process. Kohli also said the company is continuing to compete in developing cutting-edge AI agents alongside OpenAI and Anthropic.