Stanford Study Reveals AI Tools Create “Workslop,” Hurting Firms

A recent study from researchers at Stanford University and BetterUp Labs highlights significant challenges facing companies that have integrated artificial intelligence (AI) into their workflows. The research indicates that these tools, rather than enhancing productivity, are leading to the creation of subpar work, commonly referred to as “workslop.”

The ongoing survey surveyed approximately 1,150 full-time employees across various industries in the United States. The results suggest that a staggering 40 percent of these employees reported receiving low-quality AI-generated content in their work over the past month. This has raised questions about the efficacy of AI tools in professional settings, despite their touted potential to streamline tasks.

The findings echo previous research from MIT, which revealed that 95 percent of companies that adopted AI did not see any meaningful increase in revenue. A specific focus of the Stanford study was on AI coding assistants, which reportedly slowed down programmers rather than enhancing their productivity. As firms race to incorporate AI technologies, many are now reconsidering their reliance on these tools.

Kate Niederhoffer, a social psychologist and vice president at BetterUp, explained that employees often use AI to produce “low-effort, passable looking work,” which ultimately creates more tasks for their colleagues. In her article for the Harvard Business Review, Niederhoffer stated that this phenomenon is detrimental to workplace dynamics.

The term “workslop,” a play on “AI slop,” is used to define AI-generated content that appears adequate but lacks the depth and quality necessary to advance projects effectively. The researchers note that this shift in work quality requires recipients to spend additional time interpreting, correcting, or redoing the work, thereby transferring the effort from the creator to the receiver.

The survey also highlighted that 15 percent of the content received by employees was AI-generated, with 40 percent of that coming from peers and a further 16 percent from supervisors. This influx of low-quality submissions has led to a tense atmosphere, as teams often find themselves burdened with correcting AI-generated errors.

One finance sector respondent described the challenge of deciding whether to rewrite AI-generated content or assign it back to the original author. Another retail director noted the time wasted in verifying information and coordinating meetings to address discrepancies arising from AI workslop.

The study further revealed that receiving workslop negatively impacts perceptions of colleagues. Over half of respondents, specifically 54 percent, viewed their AI-using coworkers as less creative, while 42 percent considered them less trustworthy, and 37 percent perceived them as less intelligent.

Niederhoffer and her colleagues emphasize that the most concerning outcome may be the erosion of interpersonal relationships within organizations. While there are limited instances where AI can enhance productivity, the rapid adoption of these technologies often overlooks the nuanced challenges they introduce.

As businesses continue to explore the capabilities of AI, the findings from this Stanford study serve as a cautionary tale. Companies may need to reevaluate how they implement these tools to avoid the pitfalls of workslop and foster a more effective and cohesive work environment.