Emily M. Bender — Language Models and Linguistics
Gradient Dissent: Conversations on AI
00:00
How Do Models Play Into This Bias?
The word embeddings picked up on co-occurrences between the word Mexican and lots of other things that also co-occurred with negative sentiment. And then that would use as a component in this other model. So yeah, there wasn't in the underlying Yelp reviews, you know, any particular reason that the Mexican restaurants were rated lower. But it doesn't matter because the error was the system under predicting for any given restaurant.
Transcript
Play full episode
