The AI slop nobody logs: the 20-page doc that makes its problem your problem.

There's a person on almost every team now. The one who loves to experiment with AI, who forwards a twenty-page document at 11pm with a single line: "thoughts?" The doc looks immaculate. Headings, summaries, a confident executive overview. And somewhere around page four you realise they haven't actually read it either — they generated it, skimmed it, and passed it to you. The work didn't get done. The burden of doing it just moved to your desk.

It has a name now, and a price tag

Harvard Business Review gave this a label last year: "workslop" — AI-generated content that masquerades as good work but lacks the substance to move the task forward. Research from BetterUp Labs and Stanford's Social Media Lab found that around 40% of workers had received it in the past month, and each incident cost the receiver just under two hours to sort out. They put the invisible tax at roughly $186 per employee per month — over $9 million a year for a company of ten thousand people.

The same article points at the paradox underneath it: AI use at work has doubled, yet an MIT study found 95% of organisations see no measurable return on it. A lot of the gap is workslop. As HBR puts it, the polished output "shifts the burden of the work downstream, requiring the receiver to interpret, correct, or redo the work." It transfers effort from creator to receiver — and calls it productivity. Independent coverage of the research put it more bluntly: this stuff is quietly rotting companies from the inside, because the decay shows up as wasted hours and eroded trust rather than a number on a dashboard.

The skill was never producing. It was cutting.

Here's the part that gets missed. AI didn't create this problem; it just made the lazy version cheap. Anyone can now generate volume. What's scarce — what was always scarce — is relevance. The real skill is boiling a thing down to the handful of points a decision-maker can actually act on, and being willing to throw the rest away.

Producing a twenty-page document is not the work. Knowing which one paragraph in it changes the decision — that's the work.

I've argued before that deep domain expertise is the only real leverage left in an AI world, and this is the daily proof of it. Twenty years in, the value I bring to a client was never typing speed or page count. It's judgement about what to leave out. AI makes the wrong-lever move almost free — generating everything — while the right lever, compression, still needs someone who understands the topic well enough to know what's load-bearing.

Managers: read the raw forward as a signal

If you run a team, an unfiltered AI dump landing in your inbox is not neutral. It's information. It tells you whether the sender actually understands the subject, because someone who does will hand you the decision up top and keep the working underneath. Someone who doesn't will forward the wall of text and hope you'll figure out what they were supposed to.

That's not a reason to punish AI use — the opposite. It's a reason to coach the shape of it. The fix I push with teams is simple: lead with the recommendation, show the reasoning below it, attach the sources, and never send anything you haven't read yourself. Structured that way, an AI-assisted document becomes something the next person can build on instead of dig out of. It's the same reason I qualify inbound before the sales team clears an hour for it — triage and synthesis up front save everyone downstream from wasted time. And it's why I checked every row of a list the AI handed me before trusting a word of it. Verifying before you forward is the whole job.

The same tool, opposite hands

None of this is anti-AI. HBR's own researchers make the distinction between "pilots" — people who use AI with high agency to sharpen their own thinking — and "passengers," who use it to avoid thinking at all. The tool is identical. The output could not be more different. In a pilot's hands it's genuine leverage; in a passenger's, it's a cost the whole organisation absorbs without ever seeing it on a line item.

So before you forward the beautiful document, ask the honest question: have I made this clearer for the next person, or have I just made my confusion their problem? AI is a fantastic way to produce more. It's a terrible substitute for having something worth saying.

Sources & further reading

External

Harvard Business Review — AI-Generated "Workslop" Is Destroying Productivity (BetterUp Labs / Stanford Social Media Lab research; the "burden shifts to the receiver" framing, the cost figures, pilots vs. passengers). The Next Web — HBR warns AI 'workslop' is rotting companies from the inside (independent coverage of the same study).

Related posts

The death of generic AI: why deep domain expertise is the only real leverage left. Qualify inbound before your sales team clears an hour for it. The industry column was wrong on two-thirds of the list. I checked every one.