Managers are doing less of the work and more of the talking, and few feel ready for it
Artificial intelligence has taken over the spreadsheets, the scheduling and the first drafts that used to fill a manager's day. What's left, according to Anthony Belluccia, a behavioral scientist and industrial-organizational (I-O) psychologist at The Predictive Index in Boston, is the hardest part of the job, talking to people.
"AI is doing most of the doing," Belluccia said. "Instead of creating the dashboards, instead of doing the coding ourselves, which could be outsourced, I think what's left is all the talking. So the aligning, the explaining change, the giving feedback, building trust, that's what's left. AI can give you guidance for that, but it can't actually do that talking."
Belluccia said the companies that come out ahead with AI are the ones whose managers know how to have good conversations with their teams.
Confidence doesn't match reality
That shift is landing on managers who don't feel ready for it. The Predictive Index's 2026 AI and People Leadership Survey, conducted in July 2026 among nearly 400 managers and more than 200 CEOs and business leaders, found leaders and managers don't agree on how ready managers actually are for tough conversations.
Seventy-four percent of CEOs and business leaders described their managers as "very confident" handling tough conversations on their own. However, only 20% of managers said they'd rather prepare for one without a framework, a coaching guide or input from HR. Nearly half, 42%, said they know what they want to say but struggle with how to say it.
"That's a pretty big disconnect in how CEOs feel and the people who are actually having those conversations on the ground," Belluccia said. "Managers need help, they're saying."
That mismatch fits a broader pattern; a recent survey found nearly half of managers feel unprepared to lead the AI-native employees now joining their teams.
Feedback isn't received with your ears
Belluccia said managers don't need exceptional people skills to have a meaningful conversation. What matters, he said, is what he calls "feedback fit": tailoring the message to the person receiving it, rather than delivering the same message to everyone the same way.
"Feedback isn't received with your ears, it's received with your nervous system," Belluccia said. "The things that you hear could land very differently for you compared to somebody else."
He compared it to person-job fit, matching someone's skills and personality to a role. Feedback fit, he argued, works the same way.
"Research does find that feedback fit, matching the message to the receiver, actually increases job satisfaction and job performance," Belluccia said.
That tracks with a 2018 study in PLOS ONE on how personality shapes the way feedback lands, which found personality traits change how motivating the same feedback can be.
In practice, Belluccia said, that can mean something as simple as knowing whether an employee would rather hear a difficult message one on one instead of in a group, or delivered straightforwardly rather than softened.
That's the kind of context a generic AI tool doesn't have, which is part of why HRD America reported on how AI-generated feedback can fall short of what an employee actually needs to hear.
Practicing the hard conversation
Belluccia said the simplest place to start is role play. He said it's often overlooked, but it's based on what I-O psychologists call behavior modeling training, a three-step method built around watching a conversation modeled, practicing it, and then receiving feedback on that practice.
"Most people just go to step two, the practice," Belluccia said. "But most people really forget that it's really important to watch somebody else model it first. And then finally, the receiving feedback element is crucial."
He said many managers default to treating every employee the same way, partly because they're stretched thin. Belluccia pointed to Gallup's State of the Global Workplace 2026 report, which found 31% of leaders reported feeling lonely on a given day, and found leaders are less likely than the people they manage to report laughing.
"How are you going to communicate a difficult message to somebody if you haven't laughed recently, or if you feel chronically lonely at work?" Belluccia said. "It's going to be much more difficult to connect with that person, much less tailor your feedback to their work style."
That tension between overloaded managers and employees who want individual attention lines up with separate HRD America reporting that middle managers are growing more frustrated as AI reshapes their day-to-day workload.
The comfort AI can't replace
Belluccia used an analogy from psychologist Harry Harlow's well known 1950s experiment to make his point. Harlow gave infant monkeys two surrogate mothers, one made of wire that dispensed milk and one made of soft cloth that offered no nourishment at all. The monkeys spent nearly all their time clinging to the cloth mother, and ran to it for comfort whenever they were frightened, even though the wire mother was the one keeping them fed.
"Every organization is building a wire mother right now, with AI," Belluccia said. "We're building these beautiful wire mothers that disperse the milk, the answers to everything, the reports, the drafts. But whenever change gets scary at work, nobody runs to the AI dashboard. We all run to our managers."
Belluccia said the organizations that come out ahead won't be the ones with the most capable AI. They'll be the ones that keep investing in how people talk to each other and how trust gets built between them.
"People need each other," Belluccia said. "If we're not focusing on making conversations better and making people trust each other, then we're just building one wire mother, and no one feels any sense of humanity or comfort."
