124 - Jay McClelland: Deep Learning, Neural Networks, and Artificial Intelligence
Robinson's Podcast
00:00
The Photo and Pollution Argument for the Parallel Distributed Processing Model Never Capturing Human Intelligence
Chomsky argued in his rebuttal of Skinner's work in the 50s on the idea that learning could help a system discover how to use language. And he certainly helped to articulate an issue that many people have continued to focus on, namely the ability of these systems to deal with items that are completely novel. The argument was that models like our PDP models would not be able to capture these systematic rules. But it has been a challenge because neural networks tend to overfit to their understanding data.
Transcript
Play full episode
