Research: low productivity gains from AI may stem from employees using AI to produce “workslop”, or low-effort, passable work that creates more work for others
A confusing contradiction is unfolding in companies embracing generative AI tools: while workers are largely following mandates …
Harvard Business Review
Context & Ripple Effects
The research challenges the assumption that employee adoption of generative AI necessarily translates into organizational productivity: output that is merely passable can shift effort to colleagues who must assess, correct, or redo it.
Employees can generate deliverables faster, but recipients inherit review, verification, and rework when that output is low effort or insufficiently reliable.
Managers measuring activity or volume may overstate AI’s productivity benefit if they do not account for downstream quality-control work.
Second-order effects
Teams are pushed toward clearer ownership, review standards, and handoff rules, because the cost of weak AI output is borne by collaborators rather than only its author.
AI tool evaluations shift from headline output speed toward the cost per completed, usable task; products that fit existing workflows have a stronger case than tools that simply expand draft volume.
Third-order effects
If this pattern persists, workplace AI returns will depend less on broad access mandates and more on redesigning workflows around validation, accountability, and quality signals.
The broader risk is a widening gap between measured output and delivered value: organizations may automate production faster than they can govern the synthetic work entering their internal systems.
The trend: Generative AI is moving from a tool-adoption story to a workflow-quality and cost-per-useful-task challenge.
it turns out if you want to draw a square it's easier, cheaper and better to just draw a square than it is to train an algorithm on billions of squares and hope that it produces a square on the other end. [embedded post]
Wow, who would have guessed this could happen... “A confusing contradiction is unfolding in companies embracing generative AI tools: [...] Employees are using AI tools to create low-effort, passable looking work that ends up creating more work for their coworkers.” — hbr.org/2…
The article echoes what I've seen and heard from peers: AI makes it easy to produce slick but shallow work that looks plausible on the surface, yet riddled with errors or bad assumptions underneath. — Your coworkers end up having to fix or redo it. — It's workslop, and it's s…
whether it's just AI slop being AI slop, or workers “quiet quitting” and making slop, the fault is still on the employers for creating that environment. bosses need to admit to themselves that this is the best that AI can offer. cut your losses and respect your workers.
I work in civil engineering. I recently had a project in which another firm used AI to get information from the contract plans in order to design a specific thing. The AI didn't factor for some things that exist, and hallucinated others. — I spent days fixing what I should've…
Every insistence on doing something a certain way from upper management always costs less money, gets less shit done and the final result is affected in a negative way. — They force AI onto their employees because they don't actually understand the work they do. [embedded post…
“41% of workers have encountered such AI-generated output, costing nearly two hours of rework per instance and creating downstream productivity, trust, and collaboration issues” — I've edited AI-generated copy. It always needs massive rewrites. Right now, AI is only good at f…
i usually hate it when business types come up with business neologisms for the noble art of shirking. — but i think i'll let them have “workslop”. when your coworker gives you AI work pseudo-product that saves time for him (consistently a him) and takes up yours instead. — h…