podcast:: [[Lex Fridman Podcast]] guest:: [[Michael Littma]] topic:: [[AI]] discussed:: [[AI]] discussed:: [[Ted Chiang]] discussed:: [[../People/Douglas Rushkoff]]

Lex Fridman 144 - Michael Littma - Reinforcement Learning and the Future of AI

GPT-3ts out are amazing - does not necesarily mean AI is smarter than we think, but humas are dumber than we think - or much of what we do day to day is no that deep - were just going wih the flow… not that much is intentional… but enough is so we get by, we do come up with new ideas occasionally

• re-transcribe this note, it's very rough

• program or be programmed #book by [[../People/Douglas Rushkoff]]

  • In a time we all had to become literate, and the ones who were had more of asay than the ones who were not literate.

• Mention the short story by [[Ted Chiang]]

https://pca.st/episode/0c5019c9-4ef5-425c-be06-e2dc76a617c5?t=2169 Algos could manipulate us slowly, unnoticeable [[AI ethics]] Rollout term comes from backgammon

Zero shot learning Driving is such a social sctivity 1h44m

• its not exponential, its the first half of a sigmoid curve
• lex: yes, its hundreds of s curves stacked on top of each other

original file:: [[tickler-2021-01]]