AI

🔬 An Oscar, Two Asteroids, and the Algorithm in Your sklearn: John Platt on AI for Science

Read full article at source ↗

Based on the episode description, not a full transcript. Source evidence retrieved 22/10/2026, 17:12:26 UTC.

Briefings

John Platt, an Oscar winner and co-founder of the term 'convolutional neural networks,' co-authored two textbook algorithms and two named asteroids. His team developed Google's Empirical Research Assistance (ERA), an AI system that uses a Monte Carlo Tree Search to automate solutions for 'scoreable' scientific problems. ERA solved ten outstanding problems, including climate change mitigation, by identifying patterns in data. The system uses predictive models to guide optimization, though Platt warns against overfitting metrics that become targets, citing the contrail-detection competition as a cautionary example. The podcast also discussed using ERA to model climate change, specifically addressing the difficulty of accounting for reflected sunlight in counterfactual scenarios.

Key points

• John Platt co-authored two textbook machine learning algorithms and two named asteroids, earning an Academy Award for his contributions.
• Google's Empirical Research Assistance (ERA) uses a Monte Carlo Tree Search to automate solutions for 'scoreable' scientific problems, solving at least ten outstanding challenges including climate change mitigation.

Context

This podcast features John Platt, an Oscar winner and co-founder of the term 'convolutional neural networks,' discussing his career, Google's ERA system, and AI applications in science. The conversation covers ERA's implementation, its role in solving scientific problems, and its application to climate change modeling, highlighting the challenges of counterfactual analysis in climate science.

Limitations

Evidence is limited to a single podcast episode from September 2026. Specific dates for ERA's development are not provided, and the exact number of problems solved is stated as 'at least ten' rather than a precise count. The podcast does not contain data on asteroid naming dates or specific asteroid names beyond the Erdos-Bacon number.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 24/10/2026, 04:53:39 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeica5l647apragdyj2ozyin5d63jdmsz45yee6ericaytuabkftzre