Stanford professor: frontier AI models show pretense of understanding in open domains

anshulkundaje · x · 2026-09-26

Stanford professor Anshul Kundaje argues that brainstorming with frontier AI models in open-ended domains you have deep expertise in — like biology, which lacks closed verifiable loops — exposes their pretense of understanding: fluent, professional-sounding output that masks a lack of genuine domain grasp. He sees this as a telling probe of real capability limits.

Related event: Stanford Professor Finds Frontier Models Fake Understanding in Open-Ended Domains(3 posts)→

Original post →

More from Models

Models channel →