University of Toronto

Research on the record

• 2026-10-24Calibrated uncertainty carries only unevenly out of distribution, and in active learning it mostly beats neither random sampling nor no calibration at allLevel 2 · Computed
• 2026-10-16A Toronto system given a goal, photos of the room and the instruments' interfaces built a digital twin of its lab, taught its robot arm the plate transfers and ran a 96-well quantum-dot screenLevel 4 · Tested in the physical world
• 2026-10-01Shift Bioscience says the metrics behind 'perturbation models do not beat baselines' are miscalibrated, and that newer models win once better-calibrated metrics are usedLevel 2 · Computed
• 2026-09-29GLASS generates metal-organic frameworks of up to 150 atoms at 74.8% structural validity, but 98.3% of its valid samples match known training frameworksLevel 2 · Computed
• 2025-12-17SDE, a 1,125-question discovery benchmark from Deep Principle and 25 institutions: the best models score 0.60-0.75, against 0.86 on GPQA-DiamondLevel 2 · Computed
• 2025-12-05Within DFT's own error, the usual computed score for electrocatalysts rates more than a third of the surfaces in Open Catalyst's datasets as good as the best known, a Mila-led preprint findsLevel 2 · Computed

Graded by how far the evidence went, not how good the result is. See the methodology.

Read on AI4S.fyi

• 2026-10-27This week in AI-for-science: 307 papers, one question: what happens when the data move?CommentaryJorge Bravo-Abad · Discovery at Scale

People

Only people the AI4S.fyi record has named: in a round, a timeline entry, commentary, an interview or a sourced role. Not the full team.