Episode 23: Language Models, with Emily Bender

[upbeat music] Welcome back to Dreaming Against the Machine. I'm your host, Adam Becker. This week's guest is Emily Bender. She is a linguistics professor at the University of Washington, and, you know, it's rare for me to encounter someone who is better at hating on AI than I am, and who has an even hotter white hot rage about the whole thing than I do. Uh, but I think Emily actually does pretty clearly qualify in this category and, uh, that's not the only reason I brought her on. Uh, it's not even the main reason I brought her on. She's, she's a friend, she's great, and she knows a lot of really interesting stuff that I don't. And I just knew that she and I would be able to have a good conversation about not only the things that we hate about AI, but what we would like to see in a better world. So we had a good conversation, and I hope you enjoy it. [upbeat music] Emily, welcome to Dreaming Against the Machine.

I am really excited to be part of this podcast.

Oh, thank you. No, it's... I'm, I'm glad we finally made this work. We had technical difficulties and scheduling difficulties, and then more technical difficulties today, and, um, but-

It's been an adventure, but we are persisting.

Yes, exactly. It has been an adventure, but we are persisting. Also, it's good to see you. I haven't... I mean, I'm trying to think. I haven't seen you in person since-

Since Sundance.

Sundance, yeah.

Yeah.

That was in January. Jesus. Uh, and I haven't seen you on video since I was on your podcast, which was also a minute ago.

It's been a bit. That was after Sundance.

Yes, it was.

But yeah.

Yeah.

And also technical difficulties, but we persisted then too. [laughs]

Yes. Yeah, and I think we have found our way around those technical difficulties finally. Um, and we will not bore, [laughs]

[laughs]

... the listeners with the details. Uh, it's not that kind of podcast. But Emily, I'm gonna try to introduce you, and then you can fill in anything I missed, I suppose. Uh, but you are a linguist, a professor of linguistics at the University of Washington, uh, up in Seattle, and also a prominent and trenchant critic-

[laughs]

... of the tech industry and, uh, large language models in particular, and the co-author of the book, uh, The AI Con, with Alex Hanna, who we're gonna have on the podcast soon. Uh, what else? Oh, yeah, and also you said one of my favorite things about LLM, something that I tell people constantly. You said, "If nobody bothered to write it, why should I bother to read it?" Which-

Exactly

... is exactly how I feel about it too.

Yeah. [laughs] Yeah, no, I, I appreciate that intro, and it's really nice to be introduced as a linguist, which is what I am, and not as, like, a AI researcher or AI scholar, because I am not those things.

Yeah. [laughs]

Um, I like to tell people that I was minding my own business doing grammar engineering, which is a kind of computational linguistics that you can think of as, like, sentence diagramming by computer.

Hmm.

And then the whole world went crazy, and I realized that there's stuff from linguistics that is helpful for the world to know, so I started speaking out about this.

Yeah. Yeah. No, but I'm, I'm so glad that you did. I mean, I think