

TWiN 59: AI co-scientist
9 snips Apr 2, 2025
Delve into the exciting intersection of AI and neuroscience as hosts discuss Google's AI co-scientist, a tool poised to speed up research. Discover how AI is tackling the superbug challenge and revolutionizing antibiotic resistance studies. The podcast highlights AI's role in generating hypotheses for ALS and transforming traditional methods of inquiry. Conversations also address the implications of declining public funding and the growing influence of private companies in scientific research, creating a captivating dialogue on the future of science.
AI Snips
Chapters
Transcript
Episode notes
ChatGPT's Practical Use
- Tim Cheung's research associate used ChatGPT to quickly relabel video files.
- This demonstrated the practical usefulness of large language models for automating tedious tasks.
AI Co-scientist's Success
- Google's AI co-scientist generated hypotheses about antimicrobial resistance in 48 hours.
- These hypotheses matched years of research by Imperial College London, surprising the researchers.
Evaluating Hypotheses
- Evaluating scientific hypotheses is complex, unlike the clear win/loss signals in games like Go.
- AI struggles with subjective judgments of hypothesis quality.