I’m here to talk to you about AI. Again. Forgive me.

My colleague Brittany Trang recently flagged a great piece by Anna Moloney, the deputy comment and features editor at the U.K. publication City AM. Titled “There should have been an op-ed here but you filed AI slop,” the op-ed rants about the explosion of AI-written and -edited submissions Moloney has received recently.

I happened to read Moloney’s piece just after getting an email from someone asking whether a particular First Opinion essay had been AI-written. Very kindly and thoughtfully, the email writer flagged what they thought were clear signs of AI use. That would go against the First Opinion guidelines, which state that no AI-written text is permitted, even in short snippets.

The email presented a conundrum. The author of that essay attested in their author agreement that the piece was not AI-written, and the parts the email writer flagged read to me as more, well, cliché than necessarily AI.

The next step might have been an AI detector. In the City AM piece, Moloney talks about running pieces through an AI detector and then bringing the results to authors (or the comms folks working with the authors).

I have not yet taken the step of using an AI detector, for a couple of reasons.

First, I just don’t trust them. AI detectors sometimes flag 100% human-written pieces. They simply aren’t reliable enough for me to feel comfortable using them to reject an essay. This is particularly true when English isn’t the author’s first language.

I also am increasingly wondering about genuinely human-written language becoming AI-sounding. As more and more AI-written text is out there, it may subtly start to influence people’s writing. If it’s human-written, but the person has just ingested a lot of AI text, what’s an AI detector to do?

Finally, it simply feels weird to use AI to combat AI. Just this week on First Opinion, Darshak Sanghavi wrote about the AI arms race in medical billing: hospitals using AI to generate ever-higher bills and insurance using AI to fight those charges.

I’d like to keep fighting AI in First Opinion without any technological assistance — at least for as long as I can.

One more thing: Be sure to listen to this week’s “First Opinion Podcast” on the Amish and health. People often think that the Amish way of life is “dying,” but, as the two experts I spoke with explain, the Amish population doubles roughly every 20 years. The Amish approach to medicine, which has a lot in common with MAHA, could have tremendous public health implications in the future.

Recommendation of the week: In the Chronicle of Higher Education, college instructor Tyler Jagt writes, “My Students Can’t Read.” The controversial piece examines how even high-achieving students struggle to read at length — not just because of social media and smartphones, but because shifts in standardized testing emphasize reading passages rather than full books.