tech
December 8, 2025
What’s next for AlphaFold: A conversation with a Google DeepMind Nobel laureate
“I’ll be shocked if we don’t see more and more LLM impact on science,” says John Jumper.
TL;DR
- AlphaFold 2, co-led by John Jumper and Demis Hassabis, predicts protein structures with atomic accuracy, a significant advancement in biology.
- The AI system has been expanded to AlphaFold Multimer and AlphaFold 3, and its predictions have been applied to the UniProt database, covering millions of proteins.
- Scientists are using AlphaFold for various applications, including honeybee disease resistance, protein design, and identifying protein interactions for fertilization research.
- While revolutionary, AlphaFold has limitations, particularly in predicting interactions involving multiple proteins or over time, and its results require careful interpretation.
- New startups and university labs are building on AlphaFold's success to develop more specialized tools for drug discovery, aiming for even greater accuracy.
- The next frontier involves fusing AlphaFold's specialized predictive power with the broad capabilities of large language models for scientific reasoning and discovery.
Continue reading
the original article