ChatGPT Is Not Intelligent w/ Emily M. Bender
Tech Won't Save Us
00:00
The 400 Excluded Words
There is some attempt at filtering the data, because even the people who say this is just a representative sample would rather not have their dataset clogged with random text. So let's try to get those out. And when you use that list of words to filter out websites, you are gonna get rid of some porn and you are gonnaget rid of some hate websites. That's good, but it's not thorough for one thing. There's a whole bunch of words in there that actually have to do with gender identities and sexual identities which can show up on sites where people are positively speaking about these identities. If this is the future of search engines or whatnot as they're trying to pitch
Transcript
Play full episode
