

Why AI Goes Rogue: Turing Award Winner Explains Reward Failure | Andrew Barto
Andrew Barto on Rewards Gone Rogue Andrew Barto helped found reinforcement learning. He and Richard Sutton wrote the field's standard textbook, and they shared the Turing Award for that work. Reinforcement learning now trains today's reasoning models and AI agents. It also sits behind some of their strangest behavior,…

Scout bees are checking it out — divided opinions, the swarm hasn't committed.
- Platform
- 36%
- Community
- no read yet
- Value
- no read yet
Buzzmeter says it's buzzing across platforms — the Hive Vote says whether the people who actually listened liked it.
Rate this episode
Add your vote to the Hive — listeners rate every episode after they finish.
- No reviews yet — be the first.
What people are saying
“@ 39:30 I wonder if this intra-agent (adaptive control system) has anything to do with the application of the Widrow-Hoff LMS loss function using radial basis functions (as Fuzzy membership functions) on eligibility traces. If you watch som…”
“@ 22:30 I think the word that Dr. Barto is missing is the internal (dynamic, learnable, geomtric) map of the "agent" (though I call it an adaptive control system). The environment tells the state classifier where it is on the classification…”
“@ 24:10 I am guessing, but I think Dr. Barto is describing the adaptive control compensator that compensates the fuzzy, adaptive controller, in real time using as a system identification (This is a derivation of Sastry & Bodson's adaptive s…”
“@ 30:20 That is a very interesting paper that says "Taken together, these findings can be understood through quantitative theories of adaptive optimizing control."”
More from Eye On A.I.
See all episodes →
Inside Ukraine's Azov Drone R&D: The Engineer Building AI Weapons 18 km From the Front Line | Alexander Palamarchuk

Inside Ukraine's Drone War: Maj. "Phoenix" of Lasar's Group

From 10 Drones a Month to Nearly 100,000 — Inside Ukraine's Largest Drone Manufacturer | Marko Kushnir, General Cherry
One great episode in your inbox, daily — free.
One email a day, unsubscribe anytime. No spam, ever.