Insights and Compression
Juergen discusses the concept of compression in science, emphasizing how insights evolve as we simplify complex data. He illustrates this with the example of predicting patterns in falling objects, highlighting the joy of discovery. Additionally, he explores how artificial systems can be designed to seek out unsolved problems, driven by intrinsic rewards that enhance their learning and understanding of new data.In this clip
From this podcast

Lex Fridman Podcast
Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11
Related Questions