
FBL103 - Edward A. Lee: How Humans and Machines Co-Evolve
The Singularity Discussion Series
00:00
The Power of Large Language Models
The ability that the machines have to do reasoning and particularly mathematical reasoning, was really quite striking. GPT-4 also makes mistakes, but the mistakes that it makes are kind of like mistakes I might see from a really, really bright Berkeley graduate student. That's progress in roughly one to two years, which is really quite astonishing. We may have invented a machine here that can give us tremendous insights into how our own reasoning ability has emerged from the mechanics of the brain.
Transcript
Play full episode