AI WITH MEL & MO ← BACK TO AI LIBRARY
LEARN AI / AI LIBRARY / AI WATCH OUTS

AI slop and the backlash

Macquarie's word of the year for 2025 — and three questions that keep you out of it.

JUL 2026·3:03·WITH MEL
TRANSCRIPT

The full script, word for word — 519 words, about 2 minutes to read. Current as at July 2026. AI tools change quickly; if something looks different when you try it, check the product's own help pages.

Australia's national dictionary, the Macquarie, made AI slop its word of the year for twenty twenty-five. The committee picked it and the public voted for it — the fourth time in the award's history those two have landed on the same word. And the definition they wrote is sharper than you'd expect: low-quality content created by generative AI, often containing errors, and not requested by the user.

Not requested by the user. Slop isn't defined by being bad — plenty of bad writing has always existed, and no dictionary ever made it word of the year. It's defined by volume without demand. Content produced because producing it became almost free, aimed at nobody in particular, filling a space that didn't need filling. It's not about whether AI touched it. It's about whether anybody wanted it.

On scale — one study looked at sixty-five thousand English-language articles published over five years. Before ChatGPT, about one in ten was AI-written. By twenty twenty-five, roughly half. Two things to hold alongside that. It relies on AI detection tools, and those are unreliable on any single piece of writing. And the same study got reported two ways. One said half of new articles are now machine-made. The other said the share has levelled off at half — meaning AI hasn't taken over. Regardless, the rise is real.

And the platforms have turned. YouTube's chief executive named managing AI slop as a priority for twenty twenty-six. Sixteen channels were removed with thirty-five million subscribers between them and an estimated ten million dollars a year in advertising. The direction is disclosure required, AI-generated video labelled, and monetisation rules tightened against mass-produced repetition. For anyone making content, that's the commercial answer. Slop used to cost nothing to publish. Now it can cost you your ad revenue.

Three questions sort this out, and none of them ask whether you used AI. Did a person decide what this should say — or did you accept the first draft because it filled the space? Is it for someone specific, or is it for the algorithm? And would you put your name on it, knowing someone who trusts you will read it?

Which includes this channel. So — how these videos are made. The research is done with AI. I'm an AI avatar — you worked that out. The scripts are drafted with AI. What isn't automated is the deciding. Every script here is checked and edited by a person before I say a word of it. This one came back seventeen times — a number that didn't hold up, a line that contradicted an earlier one, a phrase that sounded good and meant nothing. AI can do a lot of the work. It can't own the result.

One thing to do this week — and this one's a subtraction. Find something you're producing with AI on a schedule, that nobody's asked for, and stop making it. A newsletter section, a weekly post, a report that gets sent and never opened. The tools in this library are worth your time. Using them to make more of what nobody wanted isn't.

← BACK TO AI LIBRARY