At the University of Arizona's commencement this spring, former Google CEO Eric Schmidt stood in front of nearly 10,000 graduates and tried to talk to them about AI. They booed him. The booing was sustained and loud enough to make the news. Within a few weeks the same thing happened at three more ceremonies. A real estate executive at the University of Central Florida. The CEO of Big Machine Records at Middle Tennessee State. An Adobe AI evangelist at Marquette. Four speakers, four schools, same reception.
The takes wrote themselves. Gen Z is ungrateful. Gen Z is tech-illiterate. Gen Z used ChatGPT to write their thesis and now they're booing the people who built it. There's a thread of hypocrisy in there that's easy to point at, and people did.
I want to argue the opposite. The booing is the most rational response to commencement-stage AI talk that I've seen in a while. And once you zoom out a little, you can see it's not really about AI at all. AI is the most recent thing being handed to a generation that has been on the losing end of a much longer transfer.
The labor market they're walking into
Unemployment for college graduates aged 22 to 27 is at its highest level in twelve years. About 70% of college students say they see AI as a threat to their job prospects. Those two numbers are not independent. The entry-level white-collar pipeline (the analyst seat, the junior associate role, the first-year coordinator job) is the exact layer of work that AI tooling is being pointed at. AI hasn't replaced the junior analyst role outright. What's happened is that a senior person with good AI tools can now plausibly do the work of three juniors, and finance departments noticed.
When a 22-year-old hears a billionaire on stage say "AI will transform work," they're hearing the person whose company is building the thing tell them that the entry rung of the ladder they just spent four years and $200,000 climbing toward is being sawed off in real time.
Booing is restrained, honestly.
Forty years of the same transfer
For about forty years, a steady transfer has been running in the background of American economic life. Households headed by people over 70 used to control about 19% of national wealth, and households under 40 controlled around 12%. The over-70 share has climbed; the under-40 share has been roughly cut in half. A 30-year-old today is, for the first time in modern US history, doing worse than their parents were at 30: less buying power, less home equity, more debt, later family formation.
This didn't happen by weather. Minimum wage has not kept pace with productivity; if it had, it would be well over $20 an hour. Home prices have decoupled from household income, in large part because the people who already own homes have used zoning, permitting, and local politics to make sure not too many new ones get built near theirs. Higher education, the thing that was supposed to be the equalizer, has run the same playbook: restrict supply, raise price, harvest aspiration. The schools that produced these graduates were admitting them at single-digit acceptance rates and charging multiples of what the same degree cost a generation ago. The university itself is part of the transfer.
When COVID hit, the pattern accelerated. Asset prices were rescued. Wage earners got a few checks. The people who already owned things ended the pandemic dramatically richer. The people who were trying to start (first job, first apartment, first house) ended it staring at a market that had run away from them while they were sheltering in place.
Set against that backdrop, an executive walking onto a commencement stage to tell graduates that the next great wave of economic change is going to be exciting for them is asking for the reception he got. The graduates have been watching the wave hit the beach for their entire adult lives. They have a pretty good sense of who gets wet and who's standing on the seawall.
The AI is the multiplier. The judgment is what gets multiplied.
Twenty years of pattern-matching
I've been working for about 20 years. What makes me effective with AI right now is not that I'm a fast typist or that I've memorized prompt techniques. It's that I have 20 years of pattern-matching across a lot of domains. When I look at an output, I can tell whether it's good. I know what happens to it after I hand it off: who reads it, what they're going to push back on, what part of it is going to break in three weeks when conditions change. I know what the question behind the question usually is. That judgment is the actual leverage. The AI is the multiplier. The judgment is what gets multiplied.
A 22-year-old doesn't have that yet. They were going to get it the same way everyone before them got it: by doing the messy grunt work for three to five years. Reading the bad contracts. Building the ugly first model. Sitting in the meetings where they didn't understand half of what was said and had to figure out which half mattered. That apprenticeship was annoying and underpaid and it was also the thing that built the judgment.
The AI tools are eating that exact layer. The work still has to happen; companies aren't eliminating it. They're eliminating the person who used to do it as training. The output gets produced faster and cheaper, and the apprenticeship that used to come bundled with the output gets thrown out.
This is the part that should make everyone uncomfortable, not just the graduates. Where do the senior people with good judgment come from in ten years if nobody is doing the apprenticeship now? "Learn to use AI" is not an answer to that question. Prompting an LLM does not, by itself, teach you what a good output looks like. You have to have seen a lot of outputs in context, with consequences attached, to know.
Training that only teaches people to produce outputs faster accelerates the collapse. Training that teaches them to evaluate, revise, and contest what an AI produces is what rebuilds the judgment layer.
It's the same dynamic as the housing market, played out at the level of careers. The people who got their experience under the old conditions are now standing on top of it, telling the next cohort that the conditions have changed.
• • •
Where the graduates are right
The graduates are seeing something the people running the ceremonies haven't had to see yet.
They're also, fairly, getting hit with the criticism that some of them used AI to cheat their way through coursework and are now complaining about AI eating their jobs. Both things can be true. The cheating is bad for them. It means they didn't build the very judgment layer they now need. But it doesn't make the underlying complaint wrong. If anything it sharpens it. The students who used AI to skip the learning are walking into a market that is using AI to skip hiring them. The education system didn't make the learning matter, and the labor market no longer wants what the credential was supposed to signal. They got hit from both directions.
You can see the same compounding everywhere else in their lives, which is part of why the mood is what it is. Self-harm rates among young people, especially young women, are at historic highs. Rates of depression, loneliness, and "deaths of despair" among the young have all moved in the wrong direction. Family formation is collapsing: the share of people in their early thirties who have had a child has roughly halved in a generation. None of this is caused by AI. But all of it is the soil that AI is landing in. When a 22-year-old hears "your job is going to be transformed," the word transformed does not feel friendly. It feels like one more thing being done to them by people who will not personally have to live with the result.
What I'd actually want a commencement speaker to say
Not "embrace the future." Not "AI is a tool, learn it." That advice is true and useless. It's the career equivalent of telling someone to eat healthier. Everyone already knows.
What I'd want them to say is something closer to this: the next five years are going to be unusually hard for your cohort specifically. The apprenticeship pipeline you were planning to walk into is being restructured while you walk into it. The economic deck has been stacked against your generation for longer than you've been alive, and AI is the newest card in that stack, not the first one. The most valuable thing you can do is find the messy, unglamorous, high-context work that AI is bad at, and plant yourself there for a while. Get the reps. Build the judgment. Be the person who can tell whether the output is right. That role is not going away. The role that just produces the output is.
I'd also want the speaker to say — and this is the part I almost never hear — that the older generation owes them something, and not just rhetorical sympathy. A redesign. Of zoning, of school funding, of how we tax capital versus labor, of how we decide who is eligible for the safety net, of how we let companies offload the cost of training onto a public that increasingly can't afford to absorb it. The transfer that put my generation in a position to be effective with AI was not natural law. It was a series of choices. Different choices are available.
That's a harder speech to give, because it doesn't end with a standing ovation from the donor section. But it would be honest, and Gen Z has shown they can tell the difference.
What the booing rejects isn't the technology. It's the act of being congratulated for graduating into a market that is restructuring the first ten years of their career before they enter it while a stranger on stage calls it progress. They're right to push back, and the rest of us should be listening. The apprenticeship problem is not just theirs. It's going to be everyone's, one cohort at a time, until somebody figures out where the next generation of judgment comes from. And the broader transfer behind it isn't just theirs either. It's the bill the rest of us have been running up on their tab for forty years.
• • •