Chapter
The Limitations of Human Expertise on Problem Solving
12:49 - 17:28 (04:39)
The assumption that people can articulate how and why they make their decisions is flawed. The experts' explanations for their actions are post hoc and not necessarily accurate.
Clips
Experts in various fields struggle to articulate their decision-making process, which is essential when encoding knowledge for AI.
12:49 - 14:08 (01:19)
Summary
Experts in various fields struggle to articulate their decision-making process, which is essential when encoding knowledge for AI. The crux of the issue lies in the assumption that people can clearly express how and why they make decisions, especially those based on perceptual cues.
ChapterThe Limitations of Human Expertise on Problem Solving
EpisodeLeslie Kaelbling: Reinforcement Learning, Planning, and Robotics
PodcastLex Fridman Podcast
The question of what type of reasoning should be done inside a computer is crucial in AI.
14:08 - 15:46 (01:37)
Summary
The question of what type of reasoning should be done inside a computer is crucial in AI. While humans cannot describe their own reasoning processes, various styles of reasoning need to be done inside a computer depending on the problem at hand, with abstractions playing a critical role.
ChapterThe Limitations of Human Expertise on Problem Solving
EpisodeLeslie Kaelbling: Reinforcement Learning, Planning, and Robotics
PodcastLex Fridman Podcast
The solution to reducing the state space and horizon in reasoning is through spatial and temporal abstraction, as we don't have enough information to reason at a high fidelity.
15:46 - 17:28 (01:42)
Summary
The solution to reducing the state space and horizon in reasoning is through spatial and temporal abstraction, as we don't have enough information to reason at a high fidelity.