#416 – Yann Lecun: Meta AI, Open Source, Limits of LLMs, AGI & the Future of AI
Lex Fridman Podcast
00:00
LLMs training data is can’t equate to a 4 year old
LLMs are trained on a massive amount of text, which would take an individual 170,000 years to read through at eight hours a day. Despite this enormous amount of data, it is revealed that a four-year-old child processes even more information visually in just four years of their life than the data LLMs are trained on. This suggests that sensory input provides a richer source of information than language.
Transcript
Play full episode
