The corporate AI rush is starting to look less like automation and more like extra housekeeping. Harvard Business Review says widespread use of generative AI tools can flood teams with low-quality output, raise verification costs, and slowly hollow out company knowledge – a pattern it describes as ”workslop” and ”knowledge degradation”.
The basic loop is painfully familiar: an employee asks a model for help, the result includes mistakes or fabricated details, a colleague has to clean it up, and trust in internal documents takes another hit. That is not productivity; it is admin work wearing a shiny AI badge. And as more firms chase speed and headcount savings, they may be creating the very bottleneck they meant to remove.
How ”workslop” spreads through a company
HBR’s warning is less about a single bad prompt than about scale. Once low-grade AI text starts moving through reports, memos, hiring flows, and internal reviews, the cost of checking it multiplies fast. In some organizations, the cleanup becomes so routine that people end up assigned mainly to correcting AI-generated errors – a neat reminder that ”efficiency” can be a very expensive word.
- Employees generate draft materials with public large language models.
- Those drafts include errors or ”hallucinations”.
- Colleagues spend extra time verifying and fixing the output.
- Trust in internal data and processes falls.
The bigger problem is cultural. When workers repeatedly encounter sloppy AI output, they stop trusting not just the tool but the workflow around it. That weakens corporate memory, because teams begin relying on stale or distorted procedures instead of learning from accurate records.
Why public models are causing extra friction
HBR argues that public models often produce generic, error-prone text, which means companies are paying twice: once to generate the content and again to validate it. The article suggests a narrower approach, using generative AI only where it clearly adds value, while reserving sensitive or repetitive business tasks for systems trained on internal data. That sounds less flashy than blanket deployment, but it is probably the part that survives contact with reality.
This debate is not happening in a vacuum. Across industries, the early AI playbook has shifted from ”deploy everywhere” to ”prove the ROI,” especially as companies discover that human review does not disappear just because the draft came from a model. The hiring process is feeling the strain too, with automated screening and AI-assisted applications adding another layer of noise between candidates and employers.
The next AI fight inside offices
The sharpest question now is whether companies keep treating generative AI as a default layer in every workflow, or start drawing hard boundaries around where it belongs. If HBR is right, the winners will be the firms that use AI surgically, not theatrically. The losers will be the ones that confuse more content with more competence.