Ep 13: Can a Machine Have Human Values?
As artificial intelligence gets more and more powerful, the need becomes greater to ensure that machines do the right thing. But what does that even mean? Brian Christian joins Vasant Dhar in episode 13 of Brave New World to discuss, as the title of his new book goes, the alignment problem. Useful resources: 1. Brian Christian's homepage. 2. The Alignment Problem: Machine Learning and Human Values -- Brian Christian. 3. Algorithms to Live By: The Computer Science of Human Decisions -- Brian Christian and Tom Griffiths. 4. The Most Human Human -- Brian Christian. 5. How Social Media Threatens Society -- Episode 8 of Brave New World (w Jonathan Haidt). 6. Are We Becoming a New Species? -- Episode 12 of Brave New World (w Molly Crockett). 7. The Nature of Intelligence -- Episode 7 of Brave New World (w Yann le Cunn) 8. Some Moral and Technical Consequences of Automation -- Norbert Wiener. 9.Superintelligence: Paths, Dangers, Strategies -- Nick Bostrom. 10. Human Compatible: AI and the Problem of Control -- Stuart Russell. 11. OpenAI. 12. Center for Human-Compatible AI. 13. Concrete Problems in AI Safety -- Dario Amodei, Chris Olah, Jacob Steinhardt, Paul Christiano, John Schulman, Dan Mané. 14. Machine Bias -- Julia Angwin, Jeff Larson, Surya Mattu and Lauren Kirchner. 15. Inherent Trade-Offs in the Fair Determination of Risk Scores -- Jon Kleinberg, Sendhil Mullainathan, Manish Raghavan. 16. Algorithmic Decision Making and the Cost of Fairness -- Sam Corbett-Davies, Emma Pierson, Avi Feller, Sharad Goel, Aziz Huq.. 17. Predictions Put Into Practice -- Jessica Saunders, Priscillia Hunt, John S. Hollywood 18. An Engine, Not a Camera: How Financial Models Shape Markets -- Donald MacKenzie. 19. An Anthropologist on Mars -- Oliver Sacks. 20. Deep Reinforcement Learning from Human Preferences -- Paul F Christiano, Jan Leike, Tom B Brown, Miljan Martic, Shane Legg, Dario Amadei for OpenAI & Deep Mind.