Workslop: Why the AI Productivity Boom is Failing and How to Fix It

The real cost of workslop: Insights from industry experts

The panel discussion centered on “workslop”, a term the pair coined that refers to AI-generated content that appears polished but lacks substance.

Onstage, they cited a collaborative 2025 study that revealed AI use for low-quality content ended up costing organizations US$9 million per 10,000 employees.

The research identified an “invisible tax” of US$186 per month per employee based on time spent decoding the AI output, then redoing the underlying task instead of using the AI-generated version.

Out of 1,150 full-time deskworkers surveyed, 40 said they believe they have received AI workslop in the last month, while 53 percent admitted to sending workslop.

Further, survey respondents reported significant negative emotional and relational impacts.

Approximately half of those surveyed viewed colleagues who sent workslop as less creative, capable and reliable, and most reported feeling annoyed, frustrated and confused after receiving workslop. Only 21 percent said they were impressed when they received AI-generated work.

The speakers highlighted that AI should enhance human collaboration and creativity, rather than replace human roles and discussed the potential for AI to create new job opportunities by leveraging human experience and touch.

“We have data that suggests that over a 10-year time frame, a construct called mattering, which is very much about the experience of purpose in your work and that you bring value to your work,” explained Niederhoffer after the presentation. “That’s declining probably about 6 percent from 2017 through current, and so we’re experiencing that right now… people are having more existential questions, they’re filled with fear and anxiety about what their value is.”

And while AI tools are increasingly used, disclosure remains discreet. “They’re still very private, very personal, and so there isn’t a norm yet to disclose all the time with AI,” said Niederhoffer, adding that this leads to attribution ambiguity and stalled innovation diffusion inside firms. “If you’re doing cool things with AI, you’re not sharing that with other people at your firm, then the firm doesn’t get any of that innovation, and the innovation dies.”

For Hancock and Niederhoffer, that kind of secrecy is a symptom of a deeper misalignment: companies are pouring money into models while underinvesting in the human systems that would let those tools pay off. They argue that the real rate limiter on AI isn’t capability, but whether people feel safe enough to experiment, share what works and admit what doesn’t.

“We’re probably far more premature for the actual human behavior change that needs to happen,” Niederhoffer said, pointing to leadership, culture and employee agency as the factors that explain most of the variation in AI readiness inside firms.

In that context, they see WorkSlop as what happens when humans are treated as overseers, rather than as the creative, relational core that makes augmentation valuable in the first place.

“With this shiny object that is in every single headline and every conversation, it’s too easy to forget the evergreen human behaviors and the importance of things like spending time with people, building relationships, investing in how to align with your team, coaching and mentoring. It’s really easy to substitute some of that shiny AI possibility with human stuff,” said Niederhoffer.

Hancock stopped short of calling today’s surge in AI spending a full-blown bubble, arguing instead that markets are simply ahead of where organizations are in using AI as an augmentative tool.

What troubles both him and Niederhoffer is the relentless push for AI-driven productivity that’s already fueling distress and job anxiety. “Let’s say that we’re right, and there’s actually lots of new jobs, and you know, in the end unemployment doesn’t change that much, but it’s changing, and so there will be people laid off, and I don’t know if we as a society are thinking about how we're going to support those folks.”

A lack of broader economic planning for displaced or reshaped roles is a gap that economists are starting to scrutinize more closely.

The long-term risks of automating away entry-level jobs

With companies racing to adopt AI at a high cost, investors are rewarding aggressive cost-cutting strategies that reduce the human workforce - Meta and Block both saw their share prices rise earlier this year after layoff-related announcements tied to heavy AI spending and productivity gains.

But Jan-Emmanuel De Neve, an economist and behavioral scientist from Oxford University who co-authored a paper on the behavioral dynamics of AI adoption alongside Hancock and Niederhoffer, argues that the apparent efficiency may mask serious long-term liabilities.

He draws a sharp distinction between what he calls automators, who use AI primarily to reduce headcount, and augmenters, who invest in AI to empower employees. “There’s a risk that (automating) might undercut potential for creativity, innovation and future leadership,” he said.

Cutting junior roles removes many of the youngest, most digitally fluent and creative workers from the organization, weakening innovation when firms need it most.

Beyond that, De Neve argues that aggressive AI-driven job cuts pose mounting macroeconomic risks. While the headline-grabbing announcements come from firms axing thousands of people at a time, he said most companies quietly anticipate low single‑digit cuts each year, with junior staff taking the brunt.

“I think across companies, across industries, you’ll probably see about 2 - 3 percent headcount reduction per year, adding up to that 11 percent or so in the space of three, four years,” De Neve said, citing estimates from Goldman Sachs following a survey conducted in late 2025. An updated study from the bank estimates that AI could displace 11 million US jobs, about seven percent of the workforce, over the next decade.

“This is… a bit of a bloodbath on the labor market, especially for new graduates,” he added. “We see it even at the top schools; you sense the anxiety about this, and getting into the job market has become a lot more difficult.” The fallout, in his view, is already visible in the data: “We see huge drops in youth wellbeing… below 30-year-olds are off a cliff in terms of their wellbeing.”

When people feel their prospects shrinking, he said, they are more likely to shift support toward populist parties and away from incumbent governments. For De Neve, the message to investors is that the real AI risk isn’t just model failure, but a slow-burning that eventually feeds back into markets.

In response, AI makers are floating responses like universal basic income, a four-day work week and AI-backed sovereign wealth funds.

A future of human-centered AI

Despite these challenges, Hancock and Niederhoffer expressed optimism about AI’s potential is grounded in the idea that outcomes depend on the interaction of mindsets, tech and organizational conditions.

Framing this period as a moment for human psychology to shine if AI is integrated thoughtfully, focusing on human agency and psychological safety, Hancock and Niederhoffer are optimistic.

“It’s going to be all about the person and the situation and the interaction between those two,” Niederhoffer said, even as she warned that today’s “really compelling shortcuts” make it hard for exhausted workers to resist leaning on AI.

In that spirit, Hancock argues that when AI is used to augment rather than replace, it can actually widen the scope of human work — for instance, he believes cheaper, AI-enabled tools could expand access to financial advice and increase demand for human advisors, not eliminate them.

In their view, deliberate choices about culture, trust and agency will determine whether AI ultimately amplifies human work or hollows it out.