AI Denial Is An Increasing Career RiskgettyThere's a specific kind of denial showing up across the management ranks, and it's going to be very expensive for the people holding onto it.When Leadership IQ surveyed 1,251 executives, directors, and managers, the study found that 79.5% of them personally use AI tools. These aren't skeptics just sitting on the sidelines; they're logging into ChatGPT, Copilot, and Gemini every week. And yet, when they were asked whether AI would impact their own job, 46% either said no or they weren't sure. Asked about job replacement over the next three years, and 56% didn't expect it or weren't sure it would happen.Ponder that contradiction for a second. Nearly half of the people using the most disruptive workplace technology in a generation have somehow concluded that it will reshape everyone's job but theirs.And here's the point almost everyone misses: using AI frequently is not the same as using it well.AI Readiness: The Persistence GapLEADERSHIP IQMost managers aren't doing anything with AI that would threaten a job, including their own. They're asking it to clean up an email, summarize a meeting they didn't want to sit through, or make a paragraph sound more polished. They’re just dabbling. There's no plan, no rethinking of how the work actually gets done, no attempt to offload the real substance of the role. And when your entire experience of a technology is asking it to tidy up your prose, you naturally conclude it's a helpful little assistant, not a force that could restructure your job out of existence.That shallow usage is dangerous: it's generating a false read on the technology's ceiling. The tool feels modest because most managers are using it in a modest way. The data backs this up bluntly. In Section's 2026 AI Proficiency Report, 54% of workers rated themselves proficient with AI, but when actually tested, only 10% were. Among managers, just 33% use it daily at all, and their average proficiency score is barely higher than the employees they manage. If managers were genuinely using AI well, directing it, building workflows around it, handing it real cognitive work, this wouldn't be nearly the problem it is. The proficiency gap and the perception gap turn out to be the same gap. The people barely scratching the surface of what AI can do are, unsurprisingly, the last to believe it could do their job.MORE FOR YOUAnd this belief is remarkably sticky. Leadership IQ first measured these attitudes back in 2023, when most leaders had barely touched AI. Two years later, personal usage had exploded, and the beliefs about impact hadn't moved at all. In 2023, 54% believed AI would affect their jobs. In 2025, it was still 54%. Virtually identical. All that hands-on experience, and the underlying conviction didn't budge an inch.I've started calling this the persistence gap, but a psychologist might use a blunter term: motivated reasoning. It's a defense mechanism. When a threat feels too big and too personal to absorb, the mind doesn't process it, it exempts itself. “AI is transformational. It will disrupt industries and eliminate whole roles. Obviously. But my job? My judgment, my relationships, my institutional knowledge? That's different.” The World Economic Forum found the same pattern in its own research: professionals readily acknowledge AI's importance in the abstract, then underestimate its impact on their own specific roles. Everyone believes in the flood, but nobody believes their house is in the floodplain.The Floodplain Has A Name, And It's Middle ManagementWhile managers were reassuring themselves, the data caught up.Through the first half of 2026, U.S. employers attributed 101,743 announced job cuts specifically to AI, nearly double the total for all of 2025. According to Challenger, Gray & Christmas, AI is now the single most-cited reason for job cuts in America, and has been for four consecutive months. This isn't a forecast anymore, it's a monthly tracking number.And managers aren't collateral damage in this, they're the target. Gartner projects that by the end of 2026, one in five organizations will use AI to flatten their hierarchy and eliminate more than half of their middle-management positions. MIT Sloan researchers found that at companies deploying AI at scale, a manager's span of control has stretched from the old norm of about seven reports to as many as fifteen. Manager headcount at public companies has already fallen 6.1% since 2022, outpacing the decline in both executive roles and the overall workforce.The executives doing the cutting are not being subtle about who's on the list. When Cloudflare reduced headcount this year while reporting record revenue, CEO Matthew Prince said the quiet part out loud: the vast majority of the people laid off were "measurers," his word for middle management, finance, legal, and internal auditing. Amazon's Andy Jassy has openly targeted managers who "want to put their fingerprint on everything." Companies are posting record profits and shrinking their management layers in the same breath. Growth is no longer buying managers job security.The layoffs are only the visible mechanism, too. A less visible one is doing at least as much damage. Shopify now requires teams to prove they can't do the work with AI before they're allowed to request new headcount. Bank of America told analysts it expects headcount to fall simply by not replacing people who leave, with more than 90% of its employees already using internal AI tools. Walmart announced a three-year corporate hiring freeze while its CEO said plainly that AI is "going to change literally every job." These freezes never generate a headline, but they compound quarter after quarter, and they quietly close off the roles a displaced manager might otherwise move into.Why "I Use AI Every Day" Is The Wrong ComfortThe manager reassuring himself thinks his daily AI usage is evidence he's safe. It's actually evidence of the opposite, if you understand what's being automated.Roughly 60% of a typical middle manager's week is consumed by four tasks: compiling status reports, coordinating schedules and workflows, doing first-pass reviews of other people's work, and relaying information up and down the chain. Those four things are precisely what today's AI agents do reliably and cheaply. If most of your value is coordination and information-forwarding, you have automated your own job description and handed the blueprint to your CFO.The managers whose roles are expanding are the ones doing the things AI can't, like coaching people, exercising judgment under genuine ambiguity, and making sense of chaos. The researchers have a clean dividing line for this, "coordination managers" versus "development managers." One group is being consolidated. The other is being handed larger teams. The difference between them isn't how often they touch AI. It's whether they've thought hard about what it should be doing for them.But Isn't Some Of This Just An Excuse?It's a fair objection. There's a real debate about "AI-washing," companies pinning layoffs on AI when the actual driver is cost-cutting or a correction of pandemic-era overhiring. OpenAI's Sam Altman has acknowledged it happens. Nvidia's Jensen Huang has called out executives who blame AI for cuts, saying he "really hates" the framing. So some share of these headlines is a convenient narrative rather than a clean automation story.But notice that this doesn't rescue the manager, and it may make things worse. Whether your role gets cut for a real reason or a convenient one, the rounds are landing on the same profile: the coordinator whose week is spent on the tasks AI now absorbs. And whatever the stated reason, the people who survive keep turning out to be the fluent ones. AI-washing doesn't change what you should do. It just means the label on the layoff might be wrong while the pattern underneath it is exactly right.What The Smart Ones Are Doing InsteadNone of this is a doom sermon. The same research points to a genuinely optimistic finding: the managers who lean in are pulling away from the pack, and the lever is remarkably simple.Section's data shows that employees whose managers actively expect and require AI use score 1.5 times higher in proficiency and are 2.7 times more likely to be excited about it. Yet only 7.7% of managers currently tie AI use to any performance expectation. The single most powerful move available to a manager is being pulled by fewer than one in thirteen of them. But the gap is closeable faster than most people fear. In one enterprise benchmark, only 13% of employees tested as proficient with AI agents before any training. After structured upskilling, that figure jumped to 81%. The problem was never that managers can't learn this. It's that most haven't started, and the ones in denial won't start until it's late.So if you're a manager who wants to still have a job in three years, stop exempting yourself and start doing three things.First, audit your own week honestly. What percentage of your time is coordination, status-chasing, and information relay? That's your automatable exposure. Move it to AI yourself, deliberately, before someone does it for you, and reinvest the time in coaching and judgment.Second, get genuinely proficient, not just familiar. Stop dabbling. Learn to build and direct AI workflows and agents, not just prompt a chatbot to polish an email. Assume you're in the 44% who overrate themselves until you've proven otherwise.Third, make AI use non-optional on your team. Set the expectation, model it, and tie it to how you evaluate work. It's the one behavior most correlated with a team that thrives.One last number worth tattooing somewhere visible: Gallup found that 62% of the workers laid off in this wave were people who didn't use AI. Fluency has become a job-protection variable.The comforting story, that AI will change everyone's job but yours, is the most dangerous thing a manager can believe right now. The people writing the org charts have already stopped believing it. The only question is whether managers wake up before their names are on the list.