The Last Invention EP 3: Playing the Wrong Game
122 snips
Oct 9, 2025 Explore the intriguing relationship between games and artificial intelligence. Discover how early AI victories like Deep Blue and Watson showcased both the strengths and limitations of machines. Delve into the evolution of neural networks and the pivotal moment of ImageNet that transformed deep learning. Hear about the fascinating journey of DeepMind and the Atari experiments that pushed AI limits. Finally, grapple with the trade-off between performance and understanding in the quest for true general intelligence.
AI Snips
Chapters
Books
Transcript
Episode notes
Deep Blue's Public Triumph
- In 1997 IBM's Deep Blue beat world chess champion Garry Kasparov in a widely watched match.
- The victory showcased brute-force computation but didn't create a machine that 'thought' like humans.
Games Can Mislead About Intelligence
- Deep Blue and Watson excelled at narrow tasks but lacked general intelligence or flexible reasoning.
- Observers like Demis Hassabis saw these wins as playing the wrong game for true AI progress.
From Poker To AI Safety
- Liv Boeree recounts her path from physics to professional poker and AI safety advocacy.
- She frames games as controlled environments that reveal how agents navigate varying objectives and changing environments.
