RIP Tom Stoppard, my favorite playwright. I met him, once, in a dark London alley.

Ironically, Instagram allows much more text in a single post than this platform, so I’ll just link to the story I’ve posted there: www.instagram.com/p/DRqglhljM3...

#tomstoppard
@guthrietheater.bsky.social

In honor of the great magazine editor Lewis Lapham, who died a year ago yesterday, who gave us @laphamsquarterly.bsky.social and the modern @harpers.bsky.social, here is Annie Dillard’s long-lost tribute.
(Slightly edited to make it fit Bsky. Longer version on Instagram.)
#LewisLapham #AnnieDillard

A.I. Hallucinations Are Getting Worse, Even as New Systems Become More Powerful (Gift Article) A new wave of “reasoning” systems from companies like OpenAI is producing incorrect information more often. Even the companies don’t know why.

OpenAI “found that o3—its most powerful system—hallucinated 33% of the time when running its PersonQA benchmark test…. The new o4-mini hallucinated at … 48%.
When running another test called SimpleQA, which asks more general questions, the hallucination rates for o3 and o4-mini were 51% and 79%.”

Awww no.

This is really bad #Duolingo

I don't want to be taught by AI.

I want to be taught by mother-tongue speakers of a language.

I have a 1798 streak on #Duolingo and today I'm deleting my account.

Duolingo CEO says he's getting rid of contract employees and replacing them with AI Duolingo employees can also expect their AI use to be graded in their performance reviews.

I have completely deleted my Duolingo. I was informed by one of my kids that the company has removed most of its employees and 70% of the company is run by AI.

I urge everyone to delete their app.

I lived there for years and deeply miss those mountains.

This is what happens when you let AI write the news:

@cnn.com reports that Pope Francis was “the first Latin American pontiff in nearly 1,300 years.”

I’d sure like to know about the eighth-century Latin American Pope.

Curtailing people’s ability to read widely and carefully, to locate, assess, and compare different sources for themselves, and to write in their own voice about what they find and what it means, is arguably more effective than censorship. It is also one of the most obvious effects of generative AI.

I set up this account right before #RSA2025 + #Shax2025, thinking I’d post my way through them—but between 2 conferences, 6 archives, + dozens of old and new friends, there was no time. It’s good to see so many #earlymodern historians + literary scholars here, though. Thanks for sharing your work.

That’s why I finally created an account yesterday.