All Segments
Segments Timeline
The Evolution of AlphaZero
Richard Sutton discusses the development of AlphaZero and its relationship to earlier techniques in reinforcement learning. He highlights how AlphaZero builds on concepts from the past, particularly referencing TD Gam and its success in playing backgammon. Sutton emphasizes that while AlphaZero's performance in chess was impressive, it was not a surprising leap in technology, but rather a culmination of existing methods applied effectively.
"When Alpha Zero became this viral sensation to you as somebody who literally came up with many of the techniques that were used, did it feel to you like new breakthroughs were made or does it feel lik..."
The Evolution of AlphaZero
Richard Sutton discusses the development of AlphaZero in the context of historical reinforcement learning techniques. He highlights that while AlphaZero's success may seem groundbreaking, it builds on methods established since the '90s, particularly referencing TD Gam and its predecessor, AlphaGo. Sutton emphasizes the impressive and strategic gameplay of AlphaZero, especially in chess, where it sacrifices material for positional advantages.
"When Alpha Zero became this viral sensation to you as somebody who literally came up with many of the techniques that were used, did it feel to you like new breakthroughs were made or does it feel lik..."