Rapid AI adoption left workers cleaning up a flood of errors and low-quality work
Employers around the globe are grappling with the fallout from rushing to adopt fast-changing artificial intelligence tools.
Meta has scaled back plans to axe staff and use more AI agents after the company found employees were spending more and more time cleaning up technical and security errors, Reuters reported. And after thinning its workforce, especially middle managers, in a 2023 efficiency push, the company is expanding its managerial ranks again, LinkedIn posts from workers show.
At Microsoft, which rolled out AI widelyinternally, executive vice-president Ryan Roslansky recently publicly lamented the prevalence of AI slop in his day-to-day work. After receiving “what was clearly an AI-generated doc” from a co-worker, he used his AI to summarize it and came away with no new thinking about the idea, he said. He warned his followers against falling into this potential “doom loop”, where AI does the reading and writing and leaves behind an “ocean of sameness”.
Meanwhile, Walmart store and warehouse employees are being plagued by unreasonable expectations, errors and false alerts from AI tools that assign tasks, schedules and monitor performance, Business Insider reported.
Employers are navigating a flood of errors and low-quality work that are sapping employee time, raising privacy and security risks and pushing up costs. Now, some businesses thatstarted off fast and furious on AI are pumping the brakes, implementing restrictions where experimentation was once encouraged, bringing back experienced workers and upping training efforts.
C-suite leaders say when it comes to AI, there’s a sense that “we’re driving the truck and building it at the same time, and that’s a problem,” said Johnny C Taylor Jr, president and CEO of the Society for Human Resource Management. “If we could all rewind … we would have introduced [this] in a more thoughtful, slow way.”
Despite billions of dollars that firms around the globe have invested in AI over the past few years, major returns remain to be seen: 90% of executives say AI efforts have yet to translate to productivity gains at their companies, according to an Atlanta Federal Reserveworking paper. And an Ernst & Young survey published last month found that 36% of senior AI executives at companies with at least $1bn in annual revenue reported an AI incident that led to issues including data loss, financial damage, brand damage or operational disruptions.
Firms from Microsoft to McKinsey are limiting usage for employees, as the flood of new tools and rapacious consumption weighs on bottom lines. Others, such as Palantir, Booz Allen Hamiltonand Nvidia, are restricting employee use over growing concerns about the security of their organizations’ data with AI companies, Reutersreported.
Microsoft and Walmart declined to comment. Meta, Palantir, Nvidia and Booz Allen Hamilton did not respond to requests for comment.
“We initially treated AI like a traditional technology rollout: deploy the tools, provide training, drive adoption,” Kathleen Hogan, Microsoft’s chief strategy and transformation officer, said in a recent blogpost about the company’s AI adoption journey. “We learned that access and usage do not equal transformation.”
Expertise gaps and errors
AI remains a mixed bag for managers and employees alike: more than half of employees said they “aren’t getting the AI training they need”, while most managers said they were outskilled in AI by their staff, according to an Indeed survey published in July. Just 45% of managers said they felt “equipped to effectively lead AI natives”.
Many workers feel “like they’re being asked to make use of these tools without any guidance about how they best could be deployed”, said Zack Mabel, director of research at the Georgetown University Center on Education and the Workforce. Meanwhile, managersare balancing orders to “roll this out, find productivity gains and don’t compromise on quality”, he added.
Companies are trying to address these perceived gaps by moving away from one-off webinars for broad swaths of staff and toward “sustained, role-specific training”, Taylor said.
Microsoft found that “adding agents to a broken process still leaves a broken process” and “speeding up one step just creates a longer queue at the next”, Hogan’s blogpost reads. The company found that focusing on employee training, particularly pairing up early-career engineers with more seasoned employees, boosted productivity.
One barrier to bigger AI-induced gains is the technology’s tendency to produce “plausible-looking output” with weaknesses or mistakes that are not immediately apparent, explained Jessica Reif, assistant professor of management at the Wharton School of the University of Pennsylvania.
“This makes it more likely that an error propagates through a workflow instead of being immediately corrected,” which can cause problems to snowball, said Reif.
Meanwhile, AI’s limited ability to assess human workers is still limited, which is creating tension at some companies. At Walmart, frustration is mounting among some rank-and-file store employees over inconsistent and inaccurate AI tools, Business Insider reported.
Employees described bots that set unreasonable standards for human workers while introducing a host of their own inefficiencies and errors. AI tools plotted nonsensical routes for fulfilment workers that made it harder to fill orders quickly, pumped out useless safety alerts and set unrealistic schedules that left store employees cutting corners to keep up.
These bots assume that workers are “going to be perfect every single time”, one manager told Business Insider. “We’re not going to hold people to these impossible standards.”
Walmart declined to comment.
Rising risks
Neil Elan, attorney at Stubbs Alderton & Markiles LLP, said he had observed “a growing gap” between how his corporate clients think AI is being used and how it is actually being used, which is creating issues with confidentiality, intellectual property, efficiency and employee conduct.
It’s also creating headaches for companies when workers’ use of ChatGPT spills into public view.
The night before Will Moye was set to face 3M in court over a 2020 gas explosion at a Houston manufacturing plant that killed three and damaged hundreds of homes and businesses, the attorney received a report from the company’s independent expert witness.
Skimming on his iPad, Moye noticed telltale signs of AI: strange formatting and fonts and time stamps showing that 14,500 pages were assembled in a matter of hours. Moye suspected that the author, Josh Autenrieth of Knighthawk Engineering, lacked expertise and had relied on ChatGPT, despite billing 3M $90,000 for a report arguing the company was not liable.
Autenrieth’s chat logs, which Moye requested and shared in court, show he prompted ChatGPT, “What am I looking at?” after uploading a picture of the gas detector at the heart of the case.
The jury found that 3M had failed to properly inspect the gas detection system and ordered the company to pay $61.5m to 24 homeowners. Moye said after the verdict, the jury “expressed a profound disbelief” that the company would hire an expert who would make claims “without any foundation”.
“[3M] were trying to reset the narrative,” Moye said. “And they hired an expert who’d never touched the machine.”
A 3M spokesperson said that “after independently reviewing the facts and reaching his own conclusions”, Autenrieth “used AI as a tool in preparing his report”.
Autenrieth’s legal counsel Janina Wojtkowski said ChatGPT didn’t produce his opinions or replace his “professional judgment” but assisted in formatting, structuring and drafting amid “severe deadline constraints”.
Performance boost? Not so much
As companies and workers muddle through the unknowns of early AI adoption, costs are still coming into focus, while broad productivity gains remain elusive.
In a study published in May, University of Pittsburgh researchers concluded that productivity didn’t rise after companies invested in AI and cut staff.Although executives have been overwhelmingly optimistic about AI’s potential impacts for years, “management optimism does not effectively translate into … better performance,” the paper states.
More than 112,000 job cut announcements in the US so far in 2026 have mentioned AI, making it a top reason for layoffs Gray & Christmas
Earlier this year, Meta paused its practice of tracking employees’ computer activity to train its AI Meta also recently told employees that their adoption and usage of AI tools would no longer be part of performance reviews, the Information reported
Meta did not respond to a request for comment.
Mark Ma, University of Pittsburgh professor and researcher on the AI layoff study, said leaders are focused on short-term performance, “which is why they tried to rush ahead by investing in AI and laying off workers”.
But companies’ share prices tend to dip after they announce AI-related layoffs, he noted, suggesting that “the stock market also does not think these will work out.”
“A lot of these mistakes could have been avoided,” Ma said. “Now I feel some managers have learned their lesson.”
Explore more on these topicsReuse this content
