10 October 2026
DeepMind and Biohub panel argues AlphaFold did not solve protein folding
First reported
Latent Space ran this on .
- A panel from Google DeepMind (Google's AI research lab) and Biohub (a biology research group) discussed AI in biology.
- They argued that choosing the right data and scaling approach matters more than simply making models bigger.
- They said AlphaFold, the DeepMind protein-structure predictor, has real limits and did not fully solve protein folding (how a chain of molecules becomes a 3D shape).
Our read
Only Latent Space of the 30 newsletters covered this panel. It is a follow-up to our September 5 story, where AlphaFold was reported to predict protein structures with experimental accuracy.
Earlier on this story
Written by Newskeryx from which newsletters covered this story and from our own archive.
How it was covered
Latent Spaceswyx & Alessio
Scaling laws in biology, data quality and collection, AlphaFold's limits, protein dynamics, cryo-EM, virtual cells, protein language models, interpretability, drug discovery timelines