John Platt on Google's ERA: MCTS-style Gemini tree search that auto-solves scoreable science problems
Latent Space · rss · 2026-09-23
Latent Space interviews John Platt — Academy Award winner, inventor of textbook ML algorithms, Erdős–Bacon number 6 — on Google's Empirical Research Assistance (ERA) and AI for Science.
ERA: Born from an "auto-Kaggle" idea, ERA reduces science problems to "scoreable tasks" and automates finding maximizers. Conceptually simple: Gemini maintains a tree of experiment notebooks, MCTS-style — UCB picks the most promising branch (optimistic, not greedy — even a fifth-place notebook gets mutated), proposing 10 mutations per round with history shared across branches. Performance stepped-change between Gemini 2.0 and 2.5, going from non-functional to excellent; it has already yielded at least 10 papers. On avoiding self-deception: "It's a power tool. It can slice your fingers off" — the Kaggle contrail contest was won via a half-pixel label quirk, pure Goodhart. His starter advice: "Always just fit linear regression. Just do it."
Climate: Contrails account for 1% of human-induced warming (1g of exhaust → 10kg of ice crystals acting as an infrared blanket); the fix is dropping flight levels, but counterfactual accounting stumped the team for 2+ years until ERA cracked it. Also FireSat, satellites spotting fires while they're room-sized.
Advice: Deep domain expertise above all; sometimes "hike up the mountain" and implement things yourself; hyper-optimization overfits.
More from AGI Musings
- Researcher: vibe-science won't cut it — discipline and rigor matter more than ever — divy93t · 2026-09-23
- Carnegie China launches longitudinal study tracking the global AI talent race — kevinsxu · 2026-09-23
- Stanford releases 'AI in 2026' soup-to-nuts overview of the field — CCgong · 2026-09-23
- AI Is a Compression of Millions of Years of Brain Evolution, Not an Industrial Revolution — sudoraohacker · 2026-09-23
- DeepMind Co-founder Shane Legg: My 2008 PhD Thesis Was Well Ahead of Its Time — ShaneLegg · 2026-09-23
- Vals AI RSI Index: Opus 5.5 Beats Human Baseline on One of Five Tasks, Pulling RSI Estimate to July 2027 — ResultBackground2450 · 2026-09-23