Most Workers Say Getting Good Work from AI Takes as Much Effort as Training a New Hire

Most U.S.-based workers with experience training a new employee believe getting good work from AI tools takes just as much effort as training a human, according to an October 2026 survey. The new data comes as many businesses are turning to AI for tasks traditionally handled by entry-level workers.

In April 2025, Shopify CEO Tobi Lütke issued a companywide memo requiring teams to show why AI can’t do a job before requesting new headcount. About a month later, Anthropic CEO Dario Amodei warned that AI could replace half of entry-level white-collar jobs within one to five years. The unemployment rate for recent college graduates climbed to 5.7% in June 2026, up from 4.2% four years earlier.

In October 2026, we surveyed 2,052 U.S.-based employed adults who use AI for work at least once a week. Among the 1,856 who have trained an employee, 60% said training, overseeing, checking, and fixing AI’s work takes at least as much effort as training a new employee. One in three said it takes more effort. All of this oversight work is known as “botsitting.”

It’s Not Just Casual AI Users

The survey behind our Botsitting Report shows it’s not just casual, inexperienced AI users who feel training and supervising AI takes significant time and effort: 56.1% of those who use AI multiple times per day say it’s at least as demanding as training a human.

Those with extensive AI experience are the most likely to recognize the time and effort required to produce quality output. Among those who spend more than 10 hours per week training, overseeing, checking, or fixing AI’s work, 39.3% said it takes more effort than training a person, compared to 27.8% of those spending less than an hour per week on AI oversight.

The finding also holds across different groups: 62.3% of individual contributors feel training AI takes as much or more effort, and 58.4% of managers and above agree. There’s also very little variation between generations (Gen Z, 61%; Millennials, 60.8%; Gen X, 58.4%).

What Training AI Involves

To get reliable, high-quality output, AI tools need sufficient context and training materials. Additionally, output must be checked, some prompts must be redone, and errors have to be fixed. Many workers feel those tasks are comparable to onboarding a junior employee.

Almost a third (31.6%) of survey respondents spend at least three hours per week botsitting, and 8.4% dedicate more than 10 hours each week to it.

Other studies have also confirmed botsitting’s high demands. A 2025 report from University of Chicago business professor Anders Humlum and University of Copenhagen economics professor Emilie Vestergaard found that 59% of new tasks created for workers by AI chatbots were implementation and oversight, while only 41% were “productive AI use.”

According to Max Gladysh, co-founder of AI consulting firm BotsCrew, most of the effort is front-loaded. With time, human oversight will decrease from reviewing everything to reviewing only exceptions, but oversight will never drop to zero. Gladysh says, “Companies get into trouble when they budget AI like a software install instead of like a new team member you keep managing.” 

Who Says AI is Easier?

Not everyone agrees that AI is just as much effort as training a new hire. Forty percent of respondents who have trained an employee believe AI takes less effort than training a human, and that share grows with frequency of AI use (34.2% among weekly AI users to 43.9% among those who use AI multiple times each day). Among those who spend more than 10 hours per week monitoring AI, 41% said they would get much less done without these tools.

Lakshmi Peri, founder of Naanya, a California-based firm that helps companies decide which AI projects are worth funding, says implementing AI does take more effort than training a person at first because most teams’ knowledge is scattered instead of being in one central location. “The people who say AI is harder are usually stuck in that stage,” she says. “Once it’s done, it’s easier, because the same setup can train any number of people.”

The Work Flows Upward

AI implementation requires oversight and rework when the output isn’t acceptable. The Botsitting Report is one of several studies showing that managers and leaders carry more of the rework burden, with it finding that 40.7% of managers and above spend at least three hours each week overseeing AI and fixing errors, compared to 25% of individual contributors. 

SHRM’s 2026 Navigating AI in the Workplace report found that directors and above spend six hours per week correcting AI output, compared to three hours for individual contributors. Our AI in the Workplace Report from May 2026 revealed that 57% of managers and above have had to fix work from a colleague that relied too heavily on AI, compared to 38% of individual contributors. 

As AI-related workforce reductions impact junior roles, managers and leaders are bearing the brunt of AI oversight and corrections. This shift may come with a long-term cost.

Cutting entry-level roles only postpones the training cost rather than eliminating it, according to Dr. Christiane Schroeter, a leadership strategist and professor of entrepreneurship and marketing at Cal Poly San Luis Obispo. “You’re not saving any cost; you’re simply delaying it,” she says. “Skip that junior hire now, and you’ll be recruiting a manager in five years who costs more and comes in a stranger to your business.”

Future Impact

AI oversight is quickly becoming a core job skill. Microsoft’s 2026 Work Trend Index points out that “as AI expands what people can do, it also raises the premium on good judgment.” The corresponding survey found that workers now rank quality control of AI output (50%) as the top emerging skill, ahead of critical thinking (46%).

If companies cut back on hiring and training entry-level workers, they may struggle to fill oversight roles in the future. Dr. Schroeter tells CEOs, “For every AI tool you add, ask a junior to investigate and question its output, because that is one of the best training moments for building their critical thinking instead of just accepting it.”