Opening Why AI is Harder Than We Think
Why AI is Harder Than We Think
- Length
- 53 min
- Level
- Undergraduate
- Recorded
- Santa Fe Institute2021 · Public Lecture
Chapter 1 of 11: Before the lecture
Mitchell names four fallacies behind AI's boom-and-bust cycles, from Dreyfus's first-step fallacy to wishful mnemonics.
Artificial intelligence has cycled between optimistic AI Springs and funding-starved AI Winters since the 1950s, and Melanie Mitchell of the Santa Fe Institute asks why. Speaking in the Colloquium Virtual Complexity at C3-UNAM, she starts from forecasts by Shannon, Simon and Minsky and from McDermott's 1984 warning of a winter. She then sets out four fallacies: narrow AI as a continuum to general AI, Moravec's paradox, wishful mnemonics, and intelligence as disembodied computation. Benchmark comparisons and game-playing systems show how human-like labels overstate machine understanding. Mitchell closes on embodied cognition and common sense as the dark matter of intelligence, and questions turn to the Chinese Room and complex systems science.
Artificial intelligence has cycled between optimistic AI Springs and funding-starved AI Winters since the 1950s, and Melanie Mitchell of the Santa Fe Institute asks why. Speaking in the Colloquium Virtual Complexity at C3-UNAM, she starts from forecasts by Shannon, Simon and Minsky and from McDermott's 1984 warning of a winter. She then sets out four fallacies: narrow AI as a continuum to general AI, Moravec's paradox, wishful mnemonics, and intelligence as disembodied computation. Benchmark comparisons and game-playing systems show how human-like labels overstate machine understanding. Mitchell closes on embodied cognition and common sense as the dark matter of intelligence, and questions turn to the Chinese Room and complex systems science.
Complexity: A Guided Tour
Artificial Intelligence: A Guide for Thinking Humans
Artificial Intelligence meets natural stupidity
Shortcut Learning in Deep Neural Networks
Schema networks: Zero-shot transfer with a generative causal model of intuitive physics

Your comment is kept while you sign up.Sign in