USC's Soltanolkotabi: reasoning models progress is crazy but the frontier stays jagged
jasondeanlee · x · 2026-11-12
Speaking at the Simons Institute's "Trustworthy AI: From Hallucinations to Reliable Autonomy" event, Mahdi Soltanolkotabi of USC noted that progress in reasoning models "has been really crazy recently," but the frontier is quite jagged — despite all the progress, some very weird shortcomings persist even in the latest models.
His talk, "Steering Generative Models toward Better Visual Reasoning," focuses on guiding generative models toward better visual reasoning.
More from Research
- Google and Artificial Analysis launch MKA benchmark to verify agent deliverables — OnlineInference · 2026-11-12
- AI models can prove anything, making 'mathematically trivial' ICLR reviews obsolete — nanjiang_cs · 2026-11-12
- Neuroscience heavyweights debate whether brain foundation models can actually explain the brain — patrickmineault · 2026-11-12
- JUMP Cell Painting dataset: 1.53B single-cell profiles for 115,796 compounds — manorlaboratory · 2026-11-12
- Frank Nielsen paper: α-divergences of unnormalized exponential-family densities equal partition-function Jensen divergences — FrnkNlsn · 2026-11-12
- RoboCap-10k: 10,000 hours of open egocentric robot data across 21 industries released — micoolcho · 2026-11-12