Last October, Demis Hassabis and John Jumper won the Nobel Prize in Chemistry for AlphaFold 2. The AI system solved the protein folding problem that had stumped structural biology for 50 years — predicting structures with atomic-level accuracy in hours, compared to months with traditional methods.
How Big the Impact Is
The AlphaFold Protein Structure Database now contains over 240 million predicted structures, covering every protein produced by the human body. More than 3 million researchers in over 190 countries have used it, with one-third coming from low- and middle-income countries. It has contributed to approximately 200,000 papers.
Subsequent versions continue to iterate:
- AlphaFold Multimer: predicts multi-protein complex structures
- AlphaFold 3: extends to all life molecules including DNA, RNA, and ligands
- "AlphaFold 4" (unofficial name): being developed by spin-off company Isomorphic Labs for drug discovery
DeepMind’s Dilemma
A lengthy Nature article posed a sharp question: After winning the Nobel Prize, can DeepMind still produce breakthroughs of the same magnitude?
The core issue is not whether DeepMind lacks talent or technology, but that the broader environment has changed. The LLM wave has drawn away nearly all AI funding and attention. DeepMind’s original mission is "to use AI to advance scientific discovery," but the company’s resources and talent are hard to keep entirely insulated from the LLM frenzy.
Hassabis himself is optimistic: AI-driven scientific research, he believes, will usher in a "renaissance" after a 10-to-15-year shakeout period.
Why This Matters Beyond the AI Industry
AlphaFold’s significance goes beyond "AI is impressive." For the first time, it proved to mainstream academia that AI can solve fundamental scientific problems that traditional methods cannot. The value of this precedent far exceeds any benchmark score.
If AI can achieve this in protein folding, it could produce breakthroughs of similar magnitude in materials science, drug design, climate modeling, and other fields. The question is simply: who will fund these directions that are less easy to monetize?
Sources: CocoLoop, Nature reporting