If you’ve spent the last year doomscrolling LinkedIn, Slack, or your group chat with former coworkers, you’d be forgiven for thinking artificial intelligence is moments away from kicking down the office door, stealing your badge, and CC’ing HR on the way out. The panic is everywhere. CEOs are warning of mass layoffs. Economists are issuing ominous forecasts. Even your cousin who “works in tech” suddenly has very strong opinions about prompt engineering.
And yet, plot twist, the data is not cooperating with the hysteria.
According to new research from Vanguard, the jobs most exposed to AI automation are not shrinking. They’re actually growing. Faster than before Covid. Faster than other jobs. Faster than the collective anxiety would suggest.
Yes, really.
Let’s back up. Vanguard looked at roughly 140 occupations that, on paper, should be sweating the most right now: office clerks, typists (yes, that’s still a thing), HR assistants, law clerks, data scientists, basically anyone whose job description includes “organizing,” “analyzing,” or “entering data into a system that looks like it was designed in 2004.”
These roles have a high percentage of tasks that AI could theoretically automate with little human involvement. If AI were about to wipe out jobs en masse, this is where you’d expect to see the damage first.
But instead, Vanguard found that employment in these AI-exposed jobs grew by 1.7% between mid-2023 and mid-2025. Before Covid, from 2015 to 2019, growth was only about 1%. Meanwhile, job growth for other occupations has actually slowed.
So the jobs we thought were most at risk are thriving? That’s awkward for the apocalypse narrative.
To be clear, this isn’t a victory lap. It’s more of a “maybe lower the sirens a notch” situation. Some companies are using AI to eliminate certain positions, especially at the entry level. Others are leaning on automation to make fewer people do more work. That’s real, and it’s happening.
But widespread, economy-cratering job destruction? Not yet.
At a high level, there’s simply no evidence that AI-heavy roles are seeing lower employment overall. If anything, they’re doing better than expected. One reason is painfully obvious: AI is impressive, but it’s also kind of a mess.
Anyone who’s used these tools knows the magic trick comes with a blooper reel. AI can draft a decent memo in 30 seconds and then confidently cite a court case that does not exist. It can summarize a meeting beautifully, except for the part where it gets the entire conclusion wrong. Hallucinations aren’t a bug. They’re a personality trait.
That matters. Companies are hesitant to fully hand the keys to systems that can sound brilliant while being completely incorrect. As a result, humans are still very much in the loop, especially in roles that require judgment, accountability, or the ability to say, “Wait, that doesn’t make sense.”
Here’s another surprise: wages.
If AI were quietly gutting these jobs, you’d expect pay to stagnate or fall. Instead, wages in AI-exposed occupations grew by 3.8% after inflation in the post-Covid period. Pre-Covid, real wage growth was basically flat at 0.1%. Meanwhile, jobs less exposed to AI saw much smaller wage gains.
That’s not what “robots are taking your paycheck” looks like.
This data runs directly counter to some very loud warnings from tech leaders, including predictions that AI could wipe out half of all entry-level white-collar jobs and send unemployment soaring to 20%. Those warnings aren’t coming from nobodies, and they shouldn’t be ignored. So far, though, the labor market just isn’t backing them up.
Even the Federal Reserve’s anecdotal reports, which do note companies trimming entry-level hiring thanks to AI, describe something more incremental than catastrophic. Fewer recruiting classes here. Some unfilled roles there. A manufacturer cutting office staff by 15%. Significant, yes. Society-ending, not exactly.
What about younger workers, the group everyone agrees should be most vulnerable? Vanguard checked that too, using data from millions of 401(k) participants. If AI were quietly kneecapping early careers, you’d expect fewer young people showing up in retirement plans.
That’s not happening. Enrollment among workers aged 21 to 25 remains relatively strong. That suggests that while entry-level hiring may be tougher, it’s not being obliterated by bots. At least not yet.
Even some tech executives are pushing back on the panic. One called it “the stupidest thing a company can do” to stop hiring entry-level workers because of AI. The logic is simple: if you don’t train humans now, you won’t have experienced ones later. Unless your long-term strategy involves a fully automated workforce that never quits, sues, or needs context, which, good luck.
So where does that leave us?
AI is changing how work gets done. It’s absolutely reshaping workflows. Some jobs will shrink. Some will disappear. New ones will emerge with titles no one fully understands yet. That’s always how technological shifts work.
But the idea that we’re all five years away from sitting on a beach while algorithms do everything? Please. Most of us can’t even get AI to correctly format a spreadsheet.
For now, the truth is far less dramatic and far more human. AI is powerful, flawed, and still very much a tool. The real threat isn’t that it’s replacing everyone overnight. It’s that we panic, overreact, and make bad decisions before the data actually tells us to.
And if anyone should be nervous?
Apparently, economists.
Which, frankly, feels fair.
