AI is supposed to revolutionize workforce productivity, but so far that hasn’t been the case.
One study from MIT found that a damning 95 percent of companies that gambled on integrating the tech saw no meaningful growth in revenue. Another study exploring one of its most hyped up applications, AI coding assistants, showed that programmers actually became slower when they depended on the AI tools. Meanwhile, a slew of reports tell an increasingly familiar tale of companies firing their workers to replace them with AI, only to scramble to rehire humans once they realize the tech isn’t all it was made out to be.
But why exactly is AI falling short in the workplace? In theory, shouldn’t a tool that can generate essays on the fly, spit out code, hold down a conversation on any topic, and take notes on your behalf be amazing for the economy?
A fascinating new report from researchers at Stanford and the firm BetterUp Labs explores that question. In a survey that’s still ongoing, the team examined the responses of 1,150 full-time employees in the US across multiple industries to tease out how AI content is used in the workplace and how it affects the dynamics between employees.
Their conclusion? People are using it to churn out busywork that needs to be fixed by a human with common sense, undercutting claims that it can boost productivity in the labor force.
“Employees are using AI tools to create low-effort, passable looking work that ends up creating more work for their coworkers,” wrote Kate Niederhoffer, a social psychologist and vice president of BetterUp, in a writeup for Harvard Business Review with her colleagues.
The team calls this low quality work “workslop,” in a play on “AI slop,” the slang used for describing the shoddy AI text and imagery that pollute social media. They define “workslop” as AI-generated work that “masquerades as good work, but lacks the substance to meaningfully advance a given task.”
Sure, some employees can use AI to produce polished work. But many simply hit “enter” on their prompt and pass along whatever messy output an AI spits out — because, on a very surface level, it does seem passable.
“The insidious effect of workslop is that it shifts the burden of the work downstream, requiring the receiver to interpret, correct, or redo the work,” the team wrote. “In other words, it transfers the effort from creator to receiver.”
It’s an evolution of “cognitive offloading,” the term that psychologists use to describe outsourcing your thinking to a piece of technology, be it a calculator or a search engine. AI content, however, “uses machines to offload cognitive work to another human being,” Niederhoffer and her team argue.
... continue reading