We All Hated Busy Work. Now That It's Disappearing, It Turns Out We Miss It

At companies big and small, managers are grappling with a growing problem some call “experience starvation.”

Corporate restructurings and AI use have automated away many of the repetitive tasks junior staffers used to do, like drafting reports or pulling data into spreadsheets. That’s freeing them from a lot of grunt work.

It’s also depriving them of the way entry-level workers have long gleaned the expertise to become valuable players later on—by getting in the reps.

Like a tennis player practicing serves, it’s by writing lines of code over and over that young engineers discover how to spot bugs—and learn from their mistakes. Repeatedly drawing up contracts helps beginning lawyers master underlying legal theory, just like prepping reams of corporate tax returns lets accountants know how to eventually advise clients.

“It’s important to learn from patterns and learn from repetition,” says Nadia Alaee, a human resources executive at HR-tech firm Deel. “Taking that to the next level is learning how to make really good judgment calls.”

In lieu of all that iterative work, companies are experimenting with new ways to build muscle memory. Some, like EY, have designed simulations to let junior staff practice asking clients for information or confronting a co-worker about a missed deadline. Cisco is blocking out time for workers to role play pitching to a customer, among other exercises. Cognizant launched a 12-week boot camp this year for new hires.

In short, pretend work is now standing in for real work.

It’s all part of the radical ways entry-level work is changing, and disappearing. At 5.7% in June, unemployment among recent college graduates hovers close to its highest level in five years, according to the Federal Reserve Bank of New York. Entry-level roles in fields heavily exposed to AI, like software developers and customer-service representatives, are shrinking the most, data from the Stanford Digital Economy Lab and payroll processor ADP Research show.

New hires encounter much different workplaces than the ones their managers trained in. Instead of toiling a few years on the first rung of the corporate ladder, they’re overseeing what AI has done early on—yet often without the expertise to discern quality from AI slop.

“Junior staff are really being asked to take ownership over complex decisions that require experience-based judgment that they don’t have,” says Mallory Barg Bulman, a senior director analyst at advisory firm Gartner, which coined the “experience starvation” term.

In one EY simulation, the AI takes on the role of a difficult client pushing back on a recommendation. Employees practice responding to objections, then get real-time feedback on how to finesse their responses. The idea is to gain confidence before they deal with a real client, says EY’s Simon Brown, who leads global learning.

Early evidence suggests simulated reps help. A Stanford Law School lab looking at AI in legal services tried a negotiation simulator with summer associates from the global law firm Dechert. One group didn’t have access to the tool; the other spent two hours a week on the platform, where they interacted with AI agents posing as attorneys specialized in different parts of the merger & acquisition process.

When they held a mock negotiation at the end of the summer, three of the four finalists were from the group that did the simulations, says Megan Ma, the lab’s executive director. She cautions against drawing big conclusions from a small group.

“It’s over pattern recognition after years and years and years that you start picking up maybe what the opposing counsel is thinking,” Ma says. “I don’t think that AI simulation training is going to replace altogether that observing and learning by doing,” she adds, but “you can actually pick up patterns possibly faster.”

At A & O Shearman, senior attorneys are also mentoring junior associates on how to get the best work out of AI and to know how to improve upon it, says Daren Orzechowski, a partner at the firm.

In the past, veteran lawyers would teach more junior colleagues by walking through a draft contract or memo. Now, just as often, they share how they would craft an AI prompt to draft the document.

Many early-career professionals say they, too, worry AI will weaken their ability to learn by doing. More than half—52%—of 8,800 employees recently surveyed by the IBM Institute for Business Value said their day-to-day tasks had changed over the past year. About 45% of employees surveyed said AI was already eating away at their skills.

Surgeons provide a stark case study. For decades, residents learned by assisting more senior surgeons over and over, a process often called “see one, do one, teach one,” says Matt Beane, a professor at the University of California, Santa Barbara, who studies how people work alongside intelligent machines.

Once surgeons started using robots for more of the precision work of cutting and separating tissue, though, Beane found residents got 10 to 20 times less hands-on practice during such procedures. Surgeons didn’t need help and could do everything themselves, he says.

The most successful residents made up for it by essentially breaking the rules. They went around protocol for what Beane calls “shadow learning.” They switched departments to learn from surgeons who didn’t use robotic arms or spent hundreds of extra hours on simulations.

Beane has documented how the dynamic in surgery plays out in other industries, like investment banking and startups, and he sees it across industries with AI now.

“The junior person, unfortunately, doesn’t know what it used to be like, so they don’t have much of a basis for complaint,” he says. “They just know, ‘Well, this sucks.’ ”

Decades of research show that a healthy blend of challenge, complexity and connection drive skill development, he says. Beane, who is on academic leave, is building a tool called SkillBench to help organizations improve how work gets done with AI.

Most of the 1,500 entry-level U.S. new hires at Cognizant, the technology services and consulting firm, must start at boot camp, where they learn and practice foundational skills of programming and data structures. This year, Cognizant added an AI-powered training tool called “Skillspring” that breaks down every role into individual tasks and creates a personalized learning approach around specific gaps. It may include hands-on practice or a Socratic-like back-and-forth that requires reason to get through a problem.

Rookies can learn the ropes in a way that works best for them. As AI-natives, they are open to an entirely new way of cultivating know-how and judgment, says Cognizant CEO Ravi Kumar S.

“You don’t join entry-level to do entry-level tasks,” he says. “It’s actually a positive that you’re suddenly a middle manager on day one of your first corporate job.”