Podcast: Google Fellow John Platt on AI Tractability and AI for Science
ziv_ravid · x · 2026-09-12
This week The Information Bottleneck hosts John Platt, Google Fellow and head of Applied Science at Google Research — known for SMO and classifier probability calibration. His group now tackles the physical world: contrails, orbital wildfire detection, carbon monitoring, plasma physics, and AI that writes scientific software. They'll discuss his recent Daedalus essay arguing that a problem's AI tractability tracks the availability of metrics and benchmarks — an uncomfortable constraint on where AI for science can actually go.
More from Companies & People
- Princeton to host AI Alignment & Safety launch event on Oct 19 — canondetortugas · 2026-09-12
- Soleio × Joel Le_weinstein full interview: in AI commerce, users should be visionaries — soleio · 2026-09-12
- McKinsey: Top AI Adopters See ~20% EBITDA Gains from Better Decisions, Not Headcount Cuts — mikeflache · 2026-09-12
- Survey: 68% of Finance Teams Use AI but Only 39% of CFOs Trust It Autonomously — mikeflache · 2026-09-12
- Meta's Muse AI launches without duplex audio on Android or Ray-Bans — NickPassig · 2026-09-12
- TBPN show features OpenAI, Sound Ventures and Positron guests today — johncoogan · 2026-09-12