Chapter

The Future of GPTs
listen on SpotifyListen on Youtube
46:44 - 53:04 (06:20)

The development of GPT-4 is not quite what people expected, without a clear sense of different discoveries leading up to a clearly defined general intelligence. There is a possibility that the little tweaks in GPT-4 only save a factor of three total on computing power and the same performance can be achieved by throwing three times as much compute without implementing all the little tweaks.

Clips
The development of GPT-4 is not meeting the expected milestones for achieving general intelligence and is currently on the threshold between narrow and general intelligence.
46:44 - 49:27 (02:43)
listen on SpotifyListen on Youtube
Artificial Intelligence
Summary

The development of GPT-4 is not meeting the expected milestones for achieving general intelligence and is currently on the threshold between narrow and general intelligence. There is uncertainty about whether GPT-4 can achieve true general intelligence at all.

Chapter
The Future of GPTs
Episode
#368 – Eliezer Yudkowsky: Dangers of AI and the End of Human Civilization
Podcast
Lex Fridman Podcast
The tweaks and adjustments made to AI language models such as GPT-4 may provide some temporary gains in performance, but the real progress will come from the exponential growth of compute power.
49:27 - 53:04 (03:36)
listen on SpotifyListen on Youtube
AI language models
Summary

The tweaks and adjustments made to AI language models such as GPT-4 may provide some temporary gains in performance, but the real progress will come from the exponential growth of compute power.

Chapter
The Future of GPTs
Episode
#368 – Eliezer Yudkowsky: Dangers of AI and the End of Human Civilization
Podcast
Lex Fridman Podcast