Lol, at 3:30. He basically says that Artificial Intelligence appears to be working, gets extended to other fields, then people begin realising there are issues, and attempt to squeeze out all the money before abandoning.
Really this is about the limits of symbolic/GOFAI AI. After the main talk, around 59:10 he starts describing the alternative, mentioning convolutions, Hinton, Rumelhart, distributed representation, "idealized neurons with weights".
Lol, at 3:30. He basically says that Artificial Intelligence appears to be working, gets extended to other fields, then people begin realising there are issues, and attempt to squeeze out all the money before abandoning.
Considering "Thinking fast and slow", Dreyfus seems to think that we only think fast.
There is a section on Dreyfus here which references 'Thinking Fast and Slow' https://en.wikipedia.org/wiki/Hubert_Dreyfus%27s_views_on_artificial_intelligence#Vindicated
Really this is about the limits of symbolic/GOFAI AI. After the main talk, around 59:10 he starts describing the alternative, mentioning convolutions, Hinton, Rumelhart, distributed representation, "idealized neurons with weights".
(BTW note the Futurama reference in https://en.wikipedia.org/wiki/Hubert_Dreyfus hehe)