A recent report shows that modern artificial intelligence tools are falling short of expectations when it comes to actual productivity in the workplace. According to a survey of 1,150 U.S. workers conducted by BetterUp and Stanford University’s Social Media Lab, 40 % of respondents reported receiving low-quality AI-generated work (“workslop”) in the past month that they or a colleague had to redo—leading to an estimated annual productivity loss of about $9 million for a workforce of 10,000.
Sources: Semafor, Cyber News
Key Takeaways
– AI tools are creating “workslop”: output that may look slick but lacks substance and wastes human time.
– Despite heavy investment and growth in AI hype, meaningful automation of complex or remote work remains extremely limited.
– Businesses and workers should temper expectations: AI is still more of a support tool than a replacement, and mis-application may reduce efficiency rather than boost it.
In-Depth
The promise of artificial intelligence (AI) has roared through boardrooms, venture capital decks and media headlines for years now. We hear how AI will replace rote work, free us from mundane tasks, and unlock whole new productivity frontiers. According to a growing body of evidence, though, real-world results are not keeping pace with the hype—especially when we talk about knowledge-work, remote tasks and the kinds of roles many office-based professionals hold.
A recent survey of 1,150 U.S. desk workers, carried out by BetterUp in collaboration with Stanford’s Social Media Lab and reported by Semafor, found that 40 % of respondents received AI-generated output in the past month that was deemed “workslop” (a term coined for polished-looking but low-value work). In many cases the employee or a colleague had to spend time cleaning it up or redoing it altogether. Across a firm of 10,000 employees the estimated cost: roughly $9 million in wasted productivity per year. Meanwhile, other studies highlight that widely used AI systems still struggle to automate more than a tiny fraction of remote-work tasks—Cybernews reports one benchmark at about 2.5%.
From a conservative perspective, these findings raise important caution flags. First, they suggest that corporate enthusiasm for AI should be tempered with operational discipline: rolling out AI without measuring results can easily backfire, creating more overhead than savings. Second, the hype around “AI will replace jobs” needs nuance: the technology may displace portions of a task stream or assist workers, but full automation of complex, knowledge-intensive work remains elusive. Third, for businesses especially, the practical takeaway is that AI investment needs to be tightly aligned with realistic expectations, defined workflows, and clear metrics—not simply adopted because it’s “shiny.”
For workers, this means continuing to build human-centered skills that AI finds hard to replicate: judgment, domain expertise, interpersonal ability, context understanding. If anything, the current state of AI reinforces the value of human capability rather than diminishes it. At the same time, organisations should resist the temptation to adopt AI tools wholesale without gauging impact. The initial cost savings may be alluring, but the hidden cost of “fixing” poor output or managing workflow disruption may offset those gains.
The bottom line: AI remains a promising adjunct, but not yet the supercharged productivity engine it’s often portrayed to be. Conservative organisations—those that budget realistically, pilot projects carefully, and keep humans firmly in the loop—are likely to emerge ahead of those chasing hype without grounding.