Building Machines that Learn and Think Like People

Explore the fascinating intersection of cognitive science and artificial intelligence in this 28-minute talk by Sam Gershman from Harvard University. Delve into the concept of building machines that learn and think like humans, examining key ingredients such as intuitive theories, compositionality, and learning to learn. Discover how developmental psychology, intuitive physics, and intuitive psychology contribute to human-like AI. Analyze examples like Atari games and Montezuma's Revenge to understand the challenges and progress in creating more cognitively plausible AI systems. Investigate the role of causality in machine learning and its applications in caption generation. Gain insights into the biological and cognitive plausibility of AI models and their potential real-world applications.

Introduction
Computational Neuroscience
Biology and AI
Learning from the brain
Humanlevel AI
Humanlike AI
I learned Atari games
Three questions
Five key ingredients
Intuitive theories
Examples
developmentally
intuitive physics
intuitive psychology
compositionality
benefits
Montezumas Revenge
Learning to Learn
Causality
Caption Generation
Full Circle
Psychology in AI
Applications
Biological plausibility
Cognitive plausibility