At some point, one has to wonder who really benefits from the artificial intelligence revolution.
Take jobs, for example.
AI evangelists including Tesla and SpaceX CEO Elon Musk, Microsoft co-founder Bill Gates, and OpenAI CEOSam Altman all say that the technology will have most people, to some varying degree, working less in the near future.
Last year, Gates shared on “The Tonight Show Starring Jimmy Fallon” that humans would keep some tasks for themselves, but AI would make a two- or three-day workweek possible.
Meanwhile, uber-optimist “the moon is a distraction” Musk took the sci-fi tech fantasy a few steps further. “There will be no poverty in the future” thanks to AI, a starry-eyed Musk declared on X (formerly Twitter).
On “The Joe Rogan Experience” podcast, he also shared his vision of an AI future where “everyone has abundance, everyone has excellent medical care, everyone has whatever goods and services they want,” Business Insider reported.
If these promises sound like a (bad) car salesman trying to convince you to buy that TruCoat you don’t need, then consider the source. Still, there are signs that AI is already changing the job market.
Amazon is in the midst of laying off 16,000 employees, and while CEO Andy Jassyrecently clarified that the layoffs were financially driven, not AI-related, he also said that AI wasn’t costing Amazon jobs, with an important caveat: “yet.”
Now, new research suggests that employees are the ones hurting as employers pressure them to adopt AI to make their jobs easier.
The assumption that articial intelligence always reduces work deserves deeper examination, research reveals.Photo by The Washington Post on Getty Images ·Photo by The Washington Post on Getty Images
Just like Bill Gates, AI companies are selling an efficiency revolution to enterprise partners.
For various reasons, employees have been slow to adopt AI tools, with many not seeing a great use for them, according to Gallup. But for those who have adopted the technology and are using it consistently, signs of trouble have emerged.
UC Berkeley researchers Aruna Ranganathan and Xingqi Maggie Ye conducted an eight-month study of how generative AI affected work habits at an undisclosed U.S.-based tech company with about 200 employees.
They found that employees, who received free access to enterprise generative AI tech, worked at a faster pace, took on a broader scope of tasks, and worked longer, “often without being asked to do so.”
The researchers emphasized that the company did not mandate AI use (unlike some companies that are firing employees for not adopting AI fast enough, as Moneywise reported). Nevertheless, workers said they used AI because it made “doing more” feel “possible, accessible, and in many cases intrisically rewarding.”
If this, like many of the promises surrounding AI tech, sounds too good to be true without some hidden dowside, then the rest of what the researchers found may not be very surprising.
“The changes brought about by enthusiastic AI adoption can be unsustainable, causing problems down the line,” according to Ranganathan and Ye. “Once the excitement of experimenting fades, workers can find that their workload has quietly grown and feel stretched from juggling everything that’s suddenly on their plate.”
The added workload led to burnout, cognitive fatigue, and weakened decision-making.
Ultimately, the surge in employee productivity they observed in the beginning gave way to “lower quality work, turnover, and other problems.”
From April to December last year, Ranganathan and Ye did in-person observation two days a week at the tech company.
The pair also monitored internal communication channels and conducted another 40 in-person interviews across the engineering, product, design, research, and operations areas to draw their conclusions.
They found that generative AI intensified employees’ work in three main ways.
Task expansion: Since AI can fill knowledge gaps, researchers observed employees taking on responsibilities that previously belonged to others. Product managers and designers began writing code, researchers took on engineering taks, and individuals across the organization attempted work they previously would have outsourced, deferred, or avoided entirely.
Work and non-work lines blurred: AI lowers the barrier of entry for many tasks, so this reduced friction led to workers to squeeze “small amounts of work into moments that had previously been breaks.” Researchers observed employees sending prompts during lunch, in meetings, or while waiting for a file to load. Although “these actions rarely felt like doing more work… over time they produced a workday with fewer natural pauses and a more continous involvement with work.”
More multitasking: This new cadence led workers to attempt to manage several active threads at once. Empoyees were observed “manually writing code while AI generated an alternative version, running multiple agents in parallel, or reviving long-deferred tasks because AI could ‘handle them’ in the background.”
Despite slow adoption rates, employees are increasingly using AI tools to help lighten their workloads.
Musk, Gates, and Altman are some of the richest men in the world, and the venture capital backing their respective AI plays easily eclipses their personal fortunes.
With all of that funding and backing, more employers, like Accenture, will inevitably start mandating that employees use AI. But the Berkeley researchers offer a guide to help employers make that transition as seamless as possible.
Employers should employ more intentional pauses or dedicated break periods when employees regulate their tempo. “For example, a decision pause could require, before a major decision is finalized, one counterargument and one explicit link to organizational goals — widening the attention field just enough to protect against drift,” the researchers said.
Companies with AI assistance may also benefit from sequencing. Since AI enables constant activity in the background, “organizations can benefit from norms that deliberately shape when work moves forward,” explained Ranganathan and Ye. So instead of a company making a change every time the AI spits out an ill-advised suggestion, sequencing work “encourages work to advance in coherent phases.”
AI, since it reduces the need for collaboration, may encourage employees to “silo” themselves away from the rest of the company. To combat this, human grounding is necessary, the researchers advised. Whether it’s through brief check-ins, shared reflection moments, or structured dialogue, employers should carefully avoid setting their employees adrift with AI large language models as their only companions.
