An in-depth look at how OpenAI's ChatGPT decodes input queries to produce a “reasonable continuation” of the text, the underlying dataset, neural nets, and more
It's Just Adding One Word at a Time — That ChatGPT can automatically generate something that reads even superficially … Tweets: @stephen_wolfram , @friscojosh , @scobleizer , @krishnanrohit , @chr1sa , and @garystager Tweets: Stephen Wolfram / @stephen_wolfram : What is ChatGPT doing ... and why does it work? From the lore of neural nets to what Aristotle didn't get to ... here's my version of the story: https://writings.stephenwolfram.com/ ... https://twitter.com/... Josh Hermsmeyer / @friscojosh : fantastic article https://writings.stephenwolfram.com/ ... https://twitter.com/... Robert Scoble / @scobleizer : One of the smartest entrepreneurs I've ever met gives us an understandable look into how ChatGPT “thinks.” https://twitter.com/... Rohit / @krishnanrohit : Absolutely wonderful explanation of what goes on inside a large language model. Highly recommended. https://twitter.com/... Chris Anderson / @chr1sa : We know ChatGPT “sounds human” by using probability to pick the next word in a sentence Part of being human is just this sort of mindless pattern-matching. Open your mouth & pick from a pre-trained toolbox of phrases What about the rest of sentience? Where does that come from? Gary Stager / @garystager : In case you were interested in how ChatGPT works, here's a knowledge bomb from an actual expert, Stephen Wolfram. I know he's not a rockstar futurist eduinnovator, but he is one of the most important living scientists and mathematicians. https://writings.stephenwolfram.com/ ...