Matt H
Texas
@Andrzej I agree with you that using the AI-generated clue was a sorry, inadequate substitute for human imagination. Small cultural adoptions of AI like this add up over time and will lead to the widespread adoption of this technology without any safeties for independent writers and artists. It took me 15 minutes to write a comment about it and another 5 to reply here. I think that’s a worthwhile investment of my time if it means the silent majority of folks, who are put off by AI, see that it is not found acceptable by everyone. Tech companies have invested billions to create these fancy plagiarism machines. They don’t have any place in the arts, of which I count the NYT crossword (except on particularly tricky Thursdays).
I, too, was startled by the clue for 72A. I thought maybe I was missing some memetic context for the clue. Maybe a post on Twitter or Bluesky about bad AI poetry had caught constructor Joe Rodini’s eye? Unfortunately I wasn’t able to find anything of that sort, which would serve only as a marginal excuse for an AI-generated clue. I suppose Joe is an early adopter of generative AI, that technology which hopes to do all the fun of thinking for us. If you believe, like me, that good crossword construction is an art form… (buckle in for this analogy, please…) seeing an AI-generated clue is like finding out a visual artist I trust uses AI to colorize their work. It’s like finding out a musician I respect uses AI to write their melodies. Notice I’m using the present tense, here. If Will Shortz and the rest of the team allow AI clues once, then it’s open season going forward. We’re “fortunate” that Joe included a disclaimer this time; next time we may not be so lucky. The creation of a good crossword is arcane enough that they could easily decide to use AI to speed up the process in the future. In fact they may already be doing it, and we’d never know. Next time that I see an AI-generated clue in the NYT crossword, I will cancel my sub. If the constructors (and editor) feel they can use this artificial shortcut past a human’s authentic and artistic act of creation, then I don’t feel the need to provide a human solver.
@Andrzej I think you’re exaggerating my protests in order to make me look hysterical. I’d rather you engaged with the substance of the argument instead. You’re positing that protests against the use of AI only matter if the protester also refuses to “buy the product” going forward. I concede your genius point that we do, in fact, live in a society, many aspects of which have embraced AI. Yes I have stopped using Facebook, Amazon, and Google at different points over the last decade. I hope I will not have to add NYT Games to that list in the future. Are my protests now pure enough to hold weight for you?
Adding to my previous comment, the AI clue is also clearly plagiarizing a real poem by civil rights activist and poet James Weldon Johnson. Here’s the full text for those curious: Her eyes, twin pools of mystic light, The blend of star-sheen and black night; O’er which, to sound their glamouring haze, A man might bend, and vainly gaze. Her eyes, twin pools so dark and deep, In which life’s ancient mysteries sleep; Wherein, to seek the quested goal, A man might plunge, and lose his soul.
“Storm center” … now, this constructor has an EYE for snappy clues.
Literally had everything except Rita OR_ crossed with Gal G_DOT. I thought the actresses last name was GoDOT but it’s GADOT. Ugh. Would love a moratorium on the crossword glue of ORA and GADOT for a few months. I was able to figure out the theme and I actually enjoyed the well-signaled rebuses. Great puzzle overall!
@Barry yes that’s how plagiarism works. It’s not verbatim copying. NYT had wall-to-wall coverage of the plagiarism case against Harvard president Claudine Gay a couple of years ago. Perhaps some archival reading would help you connect the dots.
@Nox THANK YOU! How hard is it to come up with a clue for EYES? This is really troubling stuff. Next we’ll see AI clues with historical inaccuracies or hallucinations.
@JohnWM still here defending the AI clue! Impressive!
@JohnWM I’ll check back in five years to see if your predictions hold up. I hope you’re right! I wouldn’t diminish the puzzle to just “a *game.*” after all, you and I have spent multiple hours commenting about the inclusion of AI in this puzzle. I think that time was worthwhile! You clearly have a vested interest in minimizing the moral, ethical, and environmental impacts of AI and I think you’ve done great work on that front today.
All 10 comments loaded