On April 15, 2022, Steven Johnson published a piece in the New York Times Magazine entitled “A.I. Is Mastering Language. Should We Trust What It Says?” I knew this piece was coming, because I had been interviewed for it, over email, a couple of weeks ago. I read it with some trepidation, because I had the sense that Johnson’s question and goals going into the article did not maintain sufficient skepticism of the claims of AI boosters. At the same time, I was also fairly confident my words weren’t going to be taken out of context because I’d been contacted by a fact checker who was verifying the quotes they intended to use.
On reading the article, my expectations were met on both counts. Ordinarily, when I encounter AI hype in media coverage of research/products that claim to be “AI”, I get inspired to write tweet threads aiming to educate folks on how to spot and thus resist such hype. (Here’s a recent example.) Johnson’s article is ~10k words long, though, and so I’ve decided to try to do the same in blog form, rather than as a tweet thread.