The False Consciousness of Large Language Model AIs

3min Snip

00:00

Play full episode

Summary

Transcript

Episode notes

"I wanted to write a book about GBD4, with GBD4. A book about the ways it and other AIs seem to poise to shape our future," he says. "Whatever train of thought I wanted to follow, whatever arbitrary pathway sparked by an old memory or a new interest, GBD4 was always completely in step with me." He asks: 'As large language model AIs are trained on the whole internet, in what ways are their outputs false in a dorno sense, where the whole is the false? To answer this question, we need to first understand what a dorno meant by this provocative claim'

Audio version of the introduction to Greylock general partner Reid Hoffman’s latest book, entitled “Impromptu: Amplifying Humanity Through AI."

Impromptu was co-written by Hoffman and ChatGPT-4, OpenAI’s latest and most powerful large language model. The book aims to serve as a sort of “travelog” of Hoffman’s experience with GPT-4, as he explores the tool’s strengths and limitations. Through hundreds of prompts, Hoffman and GPT-4 discuss the potential advantages and drawbacks of AI, argue about human nature, conceive original sci fi plots, and even try to make a few jokes. As AI is quickly advancing, Hoffman hopes his book can serve as a guide to encourage people to learn more about the technology, consider how we might use it, and ponder the complex questions about how our choices might play out in the future. You can read the text of this chapter here: https://greylock.com/greymatter/amplifying-humanity-through-ai/ ‎

Impromptu is now available:

As a free PDF:

https://lnkd.in/eZSsGYgB

On Kindle:

https://lnkd.in/eUE2ukM3

Learn more about your ad choices. Visit megaphone.fm/adchoices