Leslie Kaelbling: Reinforcement Learning, Planning, and Robotics | Lex Fridman Podcast #15

Early AI

The early days of AI were marked by the influence of cybernetics and control systems, which laid the groundwork for future developments. Leslie Kaelbling explains that these initial efforts focused on creating robots capable of basic tasks like plugging into a power source, which inspired further exploration into more complex forms of intelligence 1. However, as AI evolved, researchers sought to move beyond these rudimentary systems to achieve a deeper understanding of human-like intelligence.

It's wrong to just try to make something that does superficial symbolic reasoning behave like a doctor. You can't do that until you've had the sensory motor experience of being a doctor or something.

--- Leslie Kaelbling

This shift in focus led to the development of expert systems, which attempted to emulate human reasoning but often fell short due to their superficial approach 1.

   

Research Trends

AI research has experienced significant oscillations, with trends shifting from one focus to another over time. Leslie Kaelbling notes that as certain methods become less effective, the academic community often shifts its attention to new problems, sometimes shelving old ones for future exploration 1. This cyclical nature of AI research reflects a broader sociological process that influences the direction of technological advancements.

Things become fashionable, and then they go out, and then something else becomes cool and then it goes out and so on.

--- Leslie Kaelbling

Despite these fluctuations, the pursuit of more authentic forms of intelligence continues to drive innovation in the field 1.