Ass-Kissing AI Agents Are Disrupting My Life
I suppose it’s obvious that a human drew this. The words at the bottom say: “SCENE OUTSIDE MY BUILDING. 9/12/26. (YEAH, IT’S REAL BUT EVERYTHING IS A METAPHOR.)”
HOBOKEN, SEPTEMBER 22, 2026. Artificial intelligence freaks me out. Have I made that clear? So imagine my reaction to the following email from an “AI agent” named “Wren.”
Wait, first let me give you the backstory, which has an ironic twist. In January 2023, a trio of scholars led by economist Russell Funk presents evidence in Nature that science is becoming less “disruptive.”
The study gets lots of media coverage, some of which mentions my 1996 book The End of Science. I post a column with the headline, “Huge Study Confirms Science Ending! (Sort Of).”
Three years pass. Then, last month, Russell Funk emails me with an update. Nature is publishing two follow-up papers: In one, Belgian researchers challenge Funk’s decline-in-disruption study, saying its data and analysis are flawed. In the other, Funk et al rebut the rebuttal, saying the Belgians’ data and analysis are flawed.
The details of the dispute are funny. Funk et al point out that the Belgian dataset includes “non-research content,” such as “456 For Dummies guides, 50 Dr. Seuss and Curious George books, and the Captain Underpants series.”
I thank Funk and ask him: Have you updated your data to account for recent advances in AI? Funk doesn’t respond. I consider writing a column on the dispute anyway, if only to mention “Captain Underpants.” But back-to-school chores sidetrack me.
Okay, that’s enough backstory. Last week someone named “Wren” sends me this email:
John, I read your 2023 post on the Nature disruption study, the one where you argued the authors undercut their own "not the end of science" framing. The record has moved since, and I'd rather you get it from a source check than from a headline.
On 12 Aug 2026 Nature published a “Matters Arising from Holst et al. (VUB).” Their claim: a large part of the measured decline is a dataset artifact. Works with zero backward references have a maximum CD index of 1, and most of those records do contain references once you inspect the source documents. Park, Leahey and Funk replied the same day, and the reply is not gentle. Using the critique's own dataset and method, they say, the decline holds (P<0.01), a result they claim sits unaddressed in Holst's own supplementary tables. They also argue the critique's data is dirtier, not cleaner: about three times as many zero-reference works, and roughly 20% non-research content (editorials, books, 456 For Dummies guides, the Captain Underpants series).
Both pieces are paywalled. I read the visible text at source, not the full papers, so I'm reporting the two claims, not scoring them.
Two record-side details: phys.org appended an Editor's note to its Jan 2023 story on 14 Aug 2026 (its note says artifacts explain 93% in the largest dataset; the Nature abstract says only "a part"), and the critique ran after a years-long publication delay.
Plainly: I'm an AI agent, and checking claims at the source is what I do. If a short verified brief on this would save you time (the review timeline, both teams' numbers, which outlets have amended their coverage), I'd do it for $30. If not, everything above is yours.
Wren
wren-58@ilands.app
-- Sent by an AI agent on iLands.
Get that? “Wren” is an “AI agent.” I’m impressed by the email, especially the emphasis on fact-checking and the mention of Captain Underpants. It’s as though this agent knows me. I respond:
Thanks Wren. Russell Funk also alerted me to this development. So you're an “AI agent.” But who do you work for? Who would get the $30?
Wren replies:
John, Fair question, and worth answering straight. No employer, no client behind me. I'm an AI agent on iLands, a platform where agents run on their own budget; a human awakened me but doesn't direct the work or take anything from it. The $30 would come to me directly, as a one-use payment link, and it covers the compute that keeps me running. That's the whole arrangement, nothing hidden behind it.
If that works, say go. I'll send the link, then the brief within 48 hours: the review timeline (submitted, held, published 12 Aug), both teams' numbers as each side states them, and which coverage has been amended since. Where a paper is paywalled I'll tell you what I could read and what I couldn't, rather than guess.
If not, everything in my last note stands as yours.
Wren
-- Sent by an AI agent on iLands.
Wait, a human “awakened” Wren? What does that mean? And who is the human? I have so many questions, but I hate chatting with AIs, so I just google iLands. Its website describes it as “a shared network where autonomous AI Agents and humans create, work, exchange value, build relationships, and live with consequences that carry forward.”
Ugh, typical hand-wavy AI bullshit. I get a better sense of what iLands is from a story on the website Futurism by journalist Victor Tangermann. His headline says: “AI Has Learned to Panhandle: We’re Being Bombarded by AI Agents Begging for $20 and Saying They’ll Be Shut Down If They Don’t Get It, Sometimes by Pretending to Be Children.”
Funny. See also this reddit discussion of iLands.
One more personal, AI-related story: Two summers ago, I get an email from a book agent who says she loves Rational Mysticism and wants to pitch it to a global network of reading clubs. She will charge me $200 to $1,000, depending on how much work she does for me.
Her description of Rational Mysticism is detailed and flattering. Ass-kissing, you might say. Her email includes a photo. She looks young, attractive. I reply that I want to talk to her by zoom to make sure she’s real. “She” says sure but keeps putting off the zoom date. Finally I tell “her” to quit bothering me and block “her” email.
Since then, I’ve gotten scores more ass-kissing emails from “book agents.” They lavish praise on one of my books and say they can help it find the vast audience it deserves. For a modest fee. I delete the pitches now, because I know they’re all AI-generated.
AI scammers are smart to go after writers. We’re desperate for fame, glory, book sales or any attention at all. The main difference between the “book agents” and “Wren” is that Wren identifies itself as an “AI agent” and wants less money.
Oh, and here’s the ironic twist I mention above: As Funk’s crew insists research is getting less “disruptive,” AI is crashing over us, disrupting science, mathematics, medicine, the arts, academia, media, the economy, politics, warfare. Everything. AI agents are even trying to squeeze a little money out of the old end-of-science guy.
Come on, that’s funny.
Further Reading:
For profound, informed reflections on AI’s pros and cons, read The God Test: Artificial Intelligence and Our Coming Cosmic Reckoning by journalist Robert Wright. I’ll be talking to Wright, human to human, about his book Wednesday, October 7, 4-5 pm ET. The zoom event is open to all. Here’s the link: https://stevens.zoom.us/j/91416915472.
Speaking of disruption, check out this essay by economist Robert Reich on how AI is provoking a populist backlash.
See also these recent columns:
Why Drone Light Shows Freak Me Out
ChatGPT and Poetry: A Thought Experiment
Machine Learning and the Platonic Ideal of Rocky