"Workslop": When AI Makes More Work, Not Less

There’s a specific kind of email you’ve probably received this year. It’s well-formatted, confident, the right length, and somehow useless. The headings are there, the bullet points line up, the tone is professional, and yet after reading it twice you still don’t know what you’re supposed to do, because the substance underneath the polish isn’t actually there. You end up writing back to ask what it meant, or worse, redoing the work yourself.

Researchers have a name for that now. They call it workslop, and the term is catching on because it puts a finger on something a lot of people have felt but couldn’t quite describe: AI didn’t save anyone time, it just moved the work to you.

The illusion the productivity numbers hide

The promise of AI at work is time saved, and that promise is partly real. But a study from BetterUp Labs and Stanford’s Social Media Lab, written up in Harvard Business Review, found a leak in the pipe. Surveying U.S. employees, the researchers found that about 41% had received workslop, AI-generated content that “masquerades as good work but lacks the substance to meaningfully advance the task.” More than half admitted to sending it. And one in ten confessed that at least half of the AI-generated work they passed to colleagues was low-effort or unhelpful.

The mechanism is what makes it costly. When you send polished-but-hollow output, the work doesn’t vanish, it gets pushed downstream to whoever receives it. They have to interpret it, fix it, or redo it. Your time savings became their time sink, plus the round-trip of figuring out that the document was hollow in the first place. Across a team, that can erase the very productivity gains the AI was supposed to deliver.

It also costs you something harder to measure. The same research found that receiving workslop made people view the sender as less capable, less reliable, and less trustworthy. Among reactions, 53% reported being annoyed and over a third confused or offended. The polished document meant to make you look productive can quietly make you look careless.

Why good people produce it anyway

Workslop usually isn’t laziness. It’s a predictable result of using AI without closing the loop.

Workslop happens to conscientious people in a predictable way. You ask the tool for a draft, it produces something that looks complete, and “looks complete” is exactly the trap, because a confident, well-structured output triggers the same “done” feeling as work you actually finished. You give it a light skim, it reads fine, and you send it. The gap between looking finished and being finished is invisible at a skim and obvious to the person who has to use it. That gap is the whole problem.

It connects to a point we made in How Much Time Does AI Actually Save You?: the saved time is only real if you don’t immediately spend it creating cleanup work for someone else.

AI can produce something that looks done in seconds. Deciding whether it’s actually done is still your job.

How to not be the source

The fix isn’t to use AI less. It’s to own the last mile that AI can’t, the judgment about whether the output is actually right and actually useful.

Read it as the receiver, not the writer. Before sending anything AI helped produce, read it once asking the question the recipient will ask: what am I supposed to do with this, and is the answer actually in here? If you have to guess, so will they.

Check the substance, not the surface. AI is good at the surface, structure, fluency, formatting, and that’s precisely why a skim isn’t enough. Verify the claims, the numbers, and the recommendation. The polish is free; the correctness is your contribution.

Add the thing only you can add. The most useful version of an AI-assisted document usually has a sentence the AI couldn’t have written, your actual recommendation, the context the reader needs, the “here’s what I think we should do.” If your forwarded output has none of that, you’re a router, not a contributor, and routers are what workslop is made of.

The people who come out ahead with AI aren’t the ones who generate the most documents. They’re the ones whose name on a document still means it’s worth reading. In a workplace filling up with confident, hollow output, that reputation is getting rarer, and more valuable, by the week.