Workslop: When AI wastes time
The promises of generative AI have spread through organizations like wildfire: faster production, reduced tedious tasks, and freed-up time for strategic thinking. Yet, behind the marketing hype, a far more subtle phenomenon is undermining performance: the proliferation of superficial, AI-generated content that slows down the value chain rather than accelerating it. This information fog, which could be called "workslop," floods inboxes, muddles decision-making, and dilutes the meaning of work.
The term refers to any deliverable—memo, report, email, presentation, snippet of code—that displays a veneer of professionalism but lacks substance. At first glance, the formatting is neat, the tone seems appropriate; on second reading, the argument crumbles, the sources are missing, the logic falters, or the cited example doesn't exist. This false appearance of efficiency shifts the cognitive load onto colleagues: instead of innovating, they spend their time checking, rewriting, and clarifying.
The figures confirm this gap between enthusiasm and real value. In one year, the proportion of organizations that have automated an end-to-end process using AI has almost doubled, as has the daily use of generative tools by employees since 2023. Despite this rapid adoption, 95% of companies still see no measurable return on their technology investments — a paradox that raises questions about how these tools are integrated into daily work. Harvard Business Review
The human cost is just as significant: when an employee receives work slop, they must, on average, dedicate nearly two additional hours to unravel inaccuracies, correct errors, or sometimes start from scratch. Translated into hard cash, this represents an "invisible tax" of approximately $186 per month per employee, which the organization pays in lost productivity. IOC
From a psychosocial perspective, this proliferation of extraneous tasks creates a "second-hand cognitive load." The recipient of superficial content must use their expertise to fill in the gaps, while the sender benefits from an illusory time saving. The result: trust erodes, collaboration becomes burdensome, and the company shifts toward a corrective rather than a creative economy.
Even worse, workloping damages the internal reputation of those who use it. Recent surveys show that colleagues perceive those who provide such content as less competent, less creative, and less reliable. This vicious cycle forces everyone to double down on oversight, stifles communication, and reduces initiative. IOC
Why does this phenomenon thrive? First, because AI is often used as a substitute for critical thinking rather than as a catalyst. Under pressure from tight deadlines, some copy and paste model outputs without serious verification; others entrust large language models with tasks that would require factual rigor or regulatory compliance. The lack of shared quality standards then leaves the field open to deliverables that are "ready to send" but fundamentally empty.
Furthermore, many adoption programs neglect AI literacy. Prompt engineering, hallucination detection, and reasoned critique of responses remain rare skills. Employees lack the tools to assess the depth, coherence, and actionability of generated text. The illusion of productivity trumps the need for accuracy; the organization churns out pages but loses impact.
To reverse this trend, three levers are essential. The first lever: establishing quality safeguards. Every AI-assisted deliverable should undergo a systematic filtering process—relevance to the context, factual accuracy, original added value, clarity of next steps. Without these criteria, there is a significant risk of confusing volume with usefulness.
The second key is to provide widespread training in critical thinking about AI. Using a model like a "tireless intern" requires knowing how to ask the right question, but above all, how to examine the answer, test its limitations, and supplement it with human reflection. A culture of peer review, experimental workshops, and public feedback strengthens this collective skill.
Third lever: rethink workflows. Approvals must include a human checkpoint explicitly dedicated to substantive verification, not just formatting. Recurring corrections must be escalated to refine models or adjust use cases. In other words, treat AI not as an infallible automaton, but as a partner that can be improved, whose limitations are documented and shared.
Ultimately, the productivity we expect from AI is not the result of blind automation but of increased discipline: choosing wisely when to delegate, investing time in validation, and preserving the need for intellectual depth. Eliminating workloads means restoring the value of human thought in a machine-amplified ecosystem. This rigor does not hinder innovation; it is its essential prerequisite.
To learn more: read our article on AI in the service of sustainable work and our reflection on AI and IH in harmony.