AI IntelligenceJun 1, 2026AI Intelligence
Article
Turing Award winner Richard Sutton criticizes pure generative AI for being unable to evaluate its own outputs, making it...
Unsuitable for real scientific discovery. He argues that systems like AlphaGo, which use reinforcement learning and self-evaluation, are more promising for science.
Data Cube AI EditorialSource: The Decoder
01
Source Brief
Turing Award winner Richard Sutton criticizes pure generative AI for being unable to evaluate its own outputs, making it unsuitable for real scientific discovery. He argues that systems like AlphaGo, which use reinforcement learning and self-evaluation, are more promising for science.