US white-collar workers are increasingly burdened by 'workslot'—AI-generated content that requires heavy manual correction—despite corporate efforts to use generative AI to boost efficiency.
A recent study of 1,150 US desk workers, led by Stanford researcher Jeff Hancock, found that 40% of employees encountered workslop within a single month. On average, staff members spent 3.4 hours per month fixing these errors, a figure that could cost a 10,000-person organization roughly $8.1 million in lost productivity.
While 92% of high-level executives report that AI makes them more productive, 40% of non-managers say the technology saves them no time at all. This disconnect stems from a growing trend of companies using AI to replace human roles while mandating the remaining staff use chatbots to handle increased workloads.
The cost of automation
Ken, a copywriter for a Miami-based cybersecurity firm, said his company's decision to mandate AI use after layoffs led to a significant drop in quality. He noted that employees now spend more time resolving disagreements between different chatbots and rewriting inaccurate drafts than they did before the technology was introduced.
“Quality decreased significantly, time to produce a piece of content increased significantly and, most importantly, morale decreased,” Ken said, requesting anonymity to protect his job.
Industry pressure is driving the surge in low-quality output. Major companies including Amazon, UPS, and Target have previously implemented layoffs that employees attributed to AI's potential productivity gains. This environment forces remaining staff to produce more volume, often without adequate training.
Freelance product designer Kelly Cashin observed that colleagues frequently copy and paste chatbot messages directly into emails without reviewing them. She noted that many workers are effectively outsourcing their professional judgment to AI due to intense pressure to increase output.
Similar issues are emerging in healthcare. Philip Barrison, a researcher at the University of Michigan, observed medical staff struggling with AI-generated email replies to patients. While intended to save time, the automated responses often failed to meet the needs of clinical communication.
