AI & Human Decision Making
summary
Robotics teams build autonomous AIBO and Pepper systems that win RoboCup challenges, record a 16-0 score, and later win a world championship while requiring a team of 20+ humans for development; deep learning scaled by data delivers repeatable pattern discovery used to detect lung cancer better than doctors and to link DNA to autism; recent work shows AI solves university calculus and the three-body problem solved at speed humans cannot match, raising the question of how to measure an AI’s percent contribution toward a Nobel-level discovery.
AI-written summary · about
chapters (17)
- 0:00Intro
- 3:46Agenda
- 4:53My lab
- 6:39Robocop
- 8:49Social Robotics
- 10:34Machine Learning
- 11:49Deep Learning
- 15:31Performance and explainability
- 16:53Nobel Prizes
- 17:18Dont try
- 17:53Reveal the unexpected
- 21:54Questions in physics
- 23:30AI and data
- 25:12AI and Nobel Prizes
- 27:55repeatability
- 29:34summary
- 34:40questions
related talks
19mAI: The current state of play
54mAI: Actual Intelligence
31mArtificial Intelligence for Artificial Humanoids
14mAI Is Built to Be Applied
42mDawn of the planet of artificial intelligence
13mAI: Big Expectations
04mFrom big data to human-level artificial intelligence
36mArtificial Intelligence: The Future of Software