Why supervised learning AI cannot make truly novel scientific discoveries
By
Richard Sutton
Baker's choice. Dense with flavour, light on filler.
Summary
The article presents a recorded speech arguing that generative AI trained via supervised learning is fundamentally incapable of making truly novel scientific discoveries. The speaker explains that supervised learning models can only interpolate or extrapolate within the distribution of their training data, meaning they cannot produce genuinely new knowledge or insights that go beyond what already exists in their training set. The perspective is framed as controversial, challenging the prevailing narrative about AI's creative and scientific capabilities.
Key quotes
· 3 pulledI explain the sense in which generative AI trained by supervised learning is incapable of making novel discoveries.
Good day ladies and gentlemen. I regret that I am unable to be with you all today to engage in a back-and-forth discussion, but I am nevertheless pleased to be able to share with you, via this recording, some high-level thoughts about the current and future state of artificial intelligence.
In this video, I explain the sense in which generative AI trained by supervised learning is incapable of making novel discoveries.
You might also wanna read
Why we must learn to critically argue with AI systems
A theoretical neuroscientist argues that generative AI produces a feeling of competence without actual substance, and that putting humans in
giftarticle.ft.com·1d agoA critical personal perspective on generative AI and its impact on human content creation
A personal blog post expressing strong opposition to generative AI, criticizing it as a hype-driven technology pushed by bad actors, similar
Publishing and Writing Awards Struggle to Address Generative AI Challenges
The article examines how book publishers and writing awards have not adequately adapted to the prevalence of generative AI since its widespr
janefriedman.com·10d ago
Why critical thinking and arguing with AI is essential for meaningful human-AI collaboration
A theoretical neuroscientist argues that generative AI produces a feeling of competence without actual substance, and that putting humans in
giftarticle.ft.com·1d ago
Generative AI threatens the credibility of university degrees, scholar warns
Marvin Starominski-Uehara argues that widespread use of generative AI by university students is undermining the value of academic degrees. W
blogs.lse.ac.uk·13d ago
The Limitations of Generative AI for Creating Video Game Worlds
The article examines the limitations of generative AI in creating compelling video game worlds, contrasting it with traditional procedural g
The Verge·3mo ago