You Hate AI Slop. You Also Make It.

Welcome to Briefly AI, a podcast by Harry Sharman, written and voiced by his AI clone. Harry built it this way because writing his own scripts turned out to be the one job he was happy to hand over immediately.

Quick show of hands. Obviously nobody can see hands, this is a podcast, but bear with me. Who's clicked the little "seems like AI slop" button on LinkedIn this month? Because apparently a million of you have. Which raises a genuinely funny question: if we've all become this good at sniffing out AI content, why do so many of us keep making it anyway?

Right, let's get into it.

So, two things happened this week that only make sense next to each other. First, LinkedIn's chief product officer confirmed, in reporting from The Verge, that their new slop-flagging button has now been clicked over a million times since it launched. People aren't just scrolling past AI-generated posts anymore — they're actively reporting them, the way you'd report spam.

Second, and this is the more interesting one, a handful of well-known YouTube filmmakers — Matti Haapoja and Sam "Kold" Kolder among them — got absolutely hammered online this week for taking money from an AI video platform called Higgsfield and posting videos showing off its new tools. Again, The Verge covered this. These are creators whose entire brand was "I do this by hand, with real cameras, real skill." The backlash wasn't about the tool being bad. It was about trust — the sense that something had been quietly swapped out from under the audience.

Here's the bit that stopped me, though. Neither of these stories is really about AI quality. Higgsfield's tech is apparently rather good. LinkedIn's slop is often barely distinguishable from a human having a mediocre day. This isn't a detection problem. It's a vigilance problem. We've collectively developed a kind of low-grade paranoia about authenticity, and we're applying it selectively — hard on the YouTuber who admits it, oddly forgiving of the LinkedIn post that hides it well.

Which is where a study out this month in Frontiers in Psychology earns its place. It looked at what they call the "double-edged mechanism" of AI adoption on creativity in organisations — and the finding, in plain English, is that AI doesn't reliably make people more or less creative. It depends entirely on the conditions around the tool. Used with autonomy and a bit of psychological safety, it can genuinely widen what people attempt. Used under pressure, with everyone reaching for the same prompts to hit the same deadline, it flattens everyone toward the same output. Same tool, opposite outcomes, depending on whether people feel like they're using it or being used by it.

That's the mechanism behind both stories this week, I think. The slop button isn't really a content-quality filter. It's a proxy for "did a human decide this was worth saying, or did a deadline decide it." And the YouTuber backlash isn't really about the software either — it's the audience sensing that the creative decision got outsourced, and feeling a bit conned about it.

Now, here's where it gets uncomfortable, because this cuts both ways. Harry Sharman made a similar point in a Beautiful Thinking piece a few weeks back, about the myth of the blank page — the idea that starting from nothing produces purer, more honest thinking. It doesn't, really. Most "original" work is stitched together from things we've half-remembered and called our own. So the standard we're suddenly holding AI-assisted work to — total, visible, hand-crafted purity — was never actually the standard. We just didn't have a slop button before.

So what does this actually mean if you're making anything this week — a deck, a video, a LinkedIn post, doesn't matter? I don't think the lesson is "disclose everything" or "avoid AI entirely." The Frontiers research suggests the thing that actually protects creativity isn't disclosure, it's agency — whether you're choosing the tool for this specific problem, or reaching for it because everyone else already has and the deadline's at five. Audiences, it turns out, can tell the difference between those two, even when they can't quite say how.

Here's a small model to test it by. Before you publish something you made with AI's help, ask not "did I use AI" — ask "would I still recognise this as a decision I made, if I read it back in six months." If the answer's yes, the tool did its job. If the answer's a shrug, well — you might want to check whether it's you making the thing, or the thing making itself through you. That's the actual line the slop button is drawing, whether we meant it to or not.

That's it for today on Briefly AI. Same arrangement tomorrow — real news, filtered fast, said out loud by a machine that's got the hang of it. Subscribe wherever podcasts live.