On May 15th, 2026, Glendale Community College in Arizona held its commencement ceremony. Students walked across the stage in caps and gowns, their family sat in the audience with cameras ready, and when each graduate's name was supposed to be read aloud, the moment that ties the whole ritual together, the one thing that makes the ceremony personal, something went wrong. The names being read didn't match the people walking. Some names weren't read at all. The screen froze. The ceremony was paused.

Graduate Grace Raymer walked across the stage, sat back down, and only then realized her name had never actually been called. Her family, a self-described loud family, hadn't cheered because they never heard the cue.

Eventually, the college president, Tiffany Hernandez, took the microphone and explained. Let's have a look. So, this is here's what's happening. We're using a new AI system as our reader. Yep. Yep. So, that is That is a lesson learned for us. Um And so, a human announcer was brought in to finish the job.

All righty, everybody. I have the part of the ceremony that doesn't require AI, so as soon as we figure this out, we're going to be working next. There is a certain institutional confidence required to tell a graduate they don't get to walk at their own graduation because your robot malfunctioned. That confidence did not survive contact with the audience.

Afterwards, the college sent an apology letter to students. Grace Raymer ran that letter through an AI detector. It flagged as AI-written. So, to take stock of the situation, a college replaced the person who says your name with a machine that couldn't say your name, and then apologized with a machine that couldn't write an apology. I'm L, I have a PhD in computer science, and I analyze AI developments to understand what's actually happening beneath the hype. And right now, what's happening beneath the hype is that graduation season 2026 has accidentally become the most honest public conversation about artificial intelligence that anyone has had all year. Not because of what's being said at podiums, but because of what the audience is saying back. The Glendale incident was not an isolated event. It was the most visceral example of a pattern that has defined this commencement season across the US.

On May 8th, Gloria Caulfield, vice president of strategic alliances at the Orlando-based company Tavistock, stood before the graduating class of the University of Central Florida's College of Arts and Humanities and told them this. The rise of artificial intelligence is the next industrial revolution. Oh. Whoa. What happened? Okay. I struck a chord. May I finish?

The next day, May 9th, Scott Borchetta, CEO of Big Machine Records, addressed graduates at Middle Tennessee State University's College of Media and Entertainment. Let's have a look at him. AI is rewriting production as we sit here. I know it. Deal with it. Like I said, it's a tool. Hey, like I said, you can you can hear me now or you can pay me later. Well, deal with it is a bold rhetorical choice for a man who is actively the reason they have to deal with it in the first place. On May 15th, the same day as Glendale, former Google CEO Eric Schmidt was booed repeatedly at the University of Arizona. Let's have a look at him, too. So, today we stand on this edge of another technological transformation. One that will be larger, faster, and more consequential than what came before. It will touch every profession, every classroom, every hospital, every laboratory, every person, and every relationship you have. I know what many of you are feeling about that. I can hear you. There is a fear >> [cheering] >> There is a fear in your generation that that the future has already been written.

At Columbia University, students and faculty organized protests against the university's plan to use AI-generated voices to read graduates' names during 2026 ceremonies for the second year running. I did a full analysis of the Caulfield incident, whether AI actually is the next industrial revolution, and what that comparison reveals, which I will link at the end of this video, but the pattern itself is what I want to focus on here because something far more interesting happened on the other side of this ledger. On May 11th, Apple co-founder Steve Wozniak took the stage at Grand Valley State University in Michigan. He mentioned AI, and the crowd didn't boo, they cheered. They applauded, they laughed. You all have AI. You all have AI. Actual intelligence. My entire life in the technical world, I've been following people that were trying to figure out how to make a brain, software, hardware, synapse chips, and I was at a company where the engineers figured out how to make a brain. Takes 9 months. Every major outlet ran the same framing. Wozniak got it right, the others got it wrong. It's a messaging story. Speak to the graduates, not at them, and you'll be fine. I personally think that framing is shallow, and I think what's actually happening here is significantly more important than a lesson in public speaking.

The standard analysis of this pattern goes something like this. The speakers who got booed were tone-deaf, and Wozniak was emotionally intelligent. That is true, but it is the surface. Look at what Wozniak actually said. It would take too long to go deeply into what I think about AI, but we've been trying to create a brain. Is there a way we can duplicate a routine a trillion times and have it work like a brain? AI is one of those attempts. And then more, you should always try to think different. Don't follow the same steps as a million other people. Think, is there something I can do a little bit different? This is not anti-AI statement. This is a pro-human statement. There is a distinction, and it is not exactly trivial. What Wozniak described, duplicating a routine a trillion times, trying to create a brain, is a generally accurate lay description of neural network architecture. The entire field of deep learning is, at its foundation, an attempt to model certain properties of biological cognition in silicon. The stop-loss functions, the attention mechanisms, the transformer architecture, all of these are structures that computer scientists and researchers have understood about their own cognitive processes and attempted to replicate digitally.

Wozniak is not dismissing AI. He is describing it with unusual honesty, and then making the observation that the original, the human brain, is what produced the copy in the first place. The audience cheered because they felt seen as people, rather than future inputs in somebody else's productivity model. And that framing, from a startup analytics site, is exactly right. But here is the detail that almost nobody is discussing. On May 11th, the same weekend as Wozniak, Nvidia CEO Jensen Huang delivered a commencement speech at Carnegie Mellon University. He told graduates, "Your career starts at the beginning of the AI revolution. I cannot imagine a more exciting time to begin your life's work." He told them AI would change every industry. He told them to run, not walk towards it. Same sermon, different congregation. One was preaching to the choir, the other was preaching to the people the choir replaced. No booze, not a single one reported. The message was functionally identical to what Eric Schmidt said at Arizona, where he was booed into the ground. And the difference is that Carnegie Mellon is a widely recognized school as the birthplace of artificial intelligence. Its graduates largely understood the technology, were entering careers built on it, and had a relationship with AI that was informed rather than fearful. So, same message, different audience, opposite reaction.

The variable isn't the message, it is the context in which the message lands.

And this is where I want to be extremely careful in this video, because I am about to say something that some people will not enjoy hearing. I completely understand that the job market right now is hostile. I completely understand the concerns about being replaced. From a macroeconomic perspective, the signals are deeply concerning. I am not minimizing any of that. The data is definitely stark. According to Monster, nearly nine in 10 graduates in the class of 2026, 88%, are concerned that AI or automation could replace entry-level roles. That is up from 64% just 1 year ago. Job posting on Handshake, one of the largest platforms for entry-level roles, are down 12% below pre-pandemic levels. Goldman Sachs estimates that AI is displacing approximately 16,000 American jobs per month, disproportionately concentrated in early-level positions, data entry, customer service, administrative support. The unemployment rate for college graduates age 22 to 27 sits at 5.6% hovering near its highest level in over a decade outside the pandemic, according to the New York Federal Reserve. And a current theory report found that 37% of organizations plan to replace early career roles with AI outright. These numbers are not abstract. They describe the lived experience of people who spent 4 years and considerable debt obtaining a degree and are now sending hundreds of applications into a void. The fear is rational. The anger is definitely earned. But, and this is the part that requires nuance, please hear me out. AI is not the only variable in this equation. It's not even the primary one.

The 30-year US Treasury yield hit 5.2% this week, its highest level since July 2007, just before the global financial crisis. That surge is being driven by the energy shock from the Iran conflict, unsustainable government debt, and fears of persistent inflation. When long-term bond yields rise like this, borrowing costs rise across the entire economy, mortgages go up, business loans get more expensive, companies hire more cautiously. That affects every new graduate regardless of whether AI exists or not. Add to that the ongoing tariff wars, actual wars being fought across multiple continents, sovereign debt at extraordinary levels in virtually every major economy, and a global macroeconomic environment that is, to put it clinically, not functioning very well right now. The job market would be difficult right now even if the entire field of artificial intelligence vanished magically overnight. Perhaps 10 to 20% less difficult, but still brutal. And this is where the binary framing of pro-AI versus anti-AI becomes actively harmful because when you collapse all of that economic anxiety onto a single variable AI, you misidentify the target. And when you misidentify the target, you cannot effectively address the problem.

So, why does the binary persist? Why do people keep collapsing this into pro versus anti, cheering versus booing, instead of engaging with the actual complexity? I think the answer is uncomfortable, and it's also deeply human. Uncertainty is cognitively expensive, and certainty is very comforting. Living in nuance requires sustained mental effort. It requires holding contradictory ideas simultaneously. AI is generally useful, and it is generally displacing people. The job market is hostile, and not all of that hostility is AI's fault. Wozniak is right, and the people who boo Schmidt are also right. Holding all of that at the same time without collapsing it into a simple story is exhausting. The brain doesn't want to do it. The brain wants to conserve energy, take shortcuts, and arrive at a conclusion it could hold onto, which, if you really think about it, is also a pretty good description of a large language model. We built AI in our own image, and then got offended when it cut corners. This is why people gravitate towards binary positions, not because they're stupid, but because binary positions are efficient. They provide identity, they provide a tribe, they feel safe. They tell you whom to agree with and whom to oppose. Are you with us or against us? Do you love AI or hate it? Pick a side, and at least you're not alone. At least is gone. And this pattern is everywhere. Binary political systems, binary moral frameworks, binary debates about technology. People don't like living in the gray. They like living in the black and the white, because at least in the black and the white, the world is legible. And here is the observation I find generally remarkable. The cognitive shortcut taking that makes binary thinking so attractive, the brain's impulse to simplify, to reduce complexity, to conserve processing power is the very same cognitive limitation that makes artificial intelligence potentially valuable. Human brains have finite storage capacity, finite energy, they cannot easily perform parallel processing, they cannot process billions of data points simultaneously. The reason AI exists is because the human brain, extraordinary as it is, has boundaries. The technology is, at its foundation, an attempt to extend those boundaries, which means the same limitation that makes people unable to discuss AI properly is the limitation that makes AI worth building. That is not an irony anyone at a graduation podium is naming, but it might be the most important thing to understand about this entire debate.

Now, there is a version of everything I've just said that sounds like deal with it, which is what Scott Borchetta told those Middle Tennessee graduates. I want to be very clear about the distinction. Borchetta said deal with it from the stage of a graduation ceremony as the CEO of a record label to graduates entering an industry his company is actively reshaping with the very technology he was telling them to accept. That doesn't sound like advice to me. That is a man standing on the far side of a drawbridge telling the people on the near side that the moat is good for them. What I'm saying is different. I'm saying the anger is valid, the fear is rational, and the energy needs to be directed with precision. Booing a tech CEO at a graduation ceremony is cathartic, but it is not strategic because the fundamental problem is not that Eric Schmidt or Jensen Huang or Scott Borchetta personally believe AI is good. The problem is that the incentive structures under which these companies operate make the current trajectory virtually inevitable. Tech companies are doing exactly what their incentive structures tell them to do, maximize returns. Expecting them to voluntarily slow down AI deployment out of concern for entry-level hiring is not a serious expectation. It has never been how market incentives work and it's not going to start now. The lobbying that needs to happen, the real sustained organized political pressure is not with tech CEOs. It is with governments, regulators, legislators, the people who set the frameworks within which these companies operate. The people who determine whether there are guardrails, transition support, retraining programs, and accountability structures. The terms of deployment are where the fight is, not technology itself. Consider the historical parallel that everyone keeps reaching for, the Industrial Revolution. Yes, artisanal workers were displaced, that is true and it was painful, but the Industrial Revolution did not just happen to people. It also produced the labor movement, factory safety regulation, child labor laws, minimum wage legislation, and eventually the welfare state. The technology was not reversed, the power looms were not dismantled, but the terms on which industrial technology was deployed were thought over, negotiated, and restructured across decades of political action. The people who achieved those changes did not do it by booing factory owners at public events. They did it by organizing, by lobbying, and by forcing structural change through political systems. That is the fight that needs to happen with AI, and it is not happening at graduation podiums.

For anyone watching this who is currently job hunting, and I understand the temptation to hear all of this and feel like I'm just offering cold comfort, I want to be practical for a moment. The situation is difficult, I am not going to pretend otherwise, but given the situation, there are things you can do. Leverage AI tools to your advantage. Use them to send a thousand applications instead of 400. The law of large numbers still applies. The more you apply, the higher the probability that something lands. Pursue internships, even if they're not glamorous. Pick up experience wherever it's available, online, freelance, part-time, whatever. The first job does not need to be the perfect job. It needs to be a start. You can build from there. Be relentlessly proactive, network, engage. The graduates who will navigate this most effectively are the ones who refuse to be passive. Not because the system is fair, it's definitely not fair, but because passivity in an unfair system guarantees the worst possible outcome. And this is not inspirational advice, by the way. It is just honest advice for a genuinely difficult moment.

And here is something worth noting as a counterpoint to all of the doom. IBM, a $240 billion technology company, recently announced that it is tripling its entry-level hiring. Their chief human resources officer, Nick La Maro, put it very plainly, "The companies 3 to 5 years from now that are going to be the most successful are those companies that double down on entry-level hiring in this environment." IBM discovered that cutting young workers from the pipeline created a structural problem. They had automated the entry-level tasks, but had no one coming up through the system who understood the business well enough to do the senior work. So, they rewrote their entry-level roles to account for AI fluency, and they are hiring more young people, not fewer. IBM discovered what every organization eventually discovers, you cannot run a pipeline if you remove the pipe. That does not fix the broader landscape, but it does suggest that the relationship between AI and employment is not a simple subtraction. It is a restructuring, and the organizations that understand the distinction earliest will have a significant advantage.

And this brings me to the deeper point about what AI is actually for, beyond the headlines, beyond the job displacement figures, beyond all of the booing. Some of the problems humanity is currently facing are too large and too complex for unaided human cognition. Consider this. Most major economic institutions still use linear regression models, fundamentally straight-line approximations, to model economic systems that are non-linear, interconnected, and wildly complex. The very economic pain that these graduates are feeling is being forecast and managed with tools that are by design inadequate for the complexity of the system they are trying to describe. A non-linear world modeled with linear tools produces linear surprises, and then the people act shocked. Climate modeling, drug discovery, protein folding, properly simulating economic systems with enough variables to actually capture how they behave, these are problems where human brains, finite in storage, finite in energy, unable to perform parallel processing at scale, generally need computational assistance. Not because human intelligence is inadequate, but because the problems have grown larger than any individual mind can hold. AI was not built to make human brains obsolete. It was built because human brains, for all their extraordinary capability, have limits. And some of the challenges facing this generation, the macroeconomic instability, the climate trajectory, the complexity of global systems, are challenges that sit beyond those limits.

The question then is not whether AI should exist. That question is already settled by the weight of investment, infrastructure, and institutional commitment behind it. 17 multi-billion-dollar corporations funded to the ceiling and incentivized at every level are not going to stop building because graduates booed at a podium. That is simply not realistic. What is realistic is demanding that the deployment of this technology happens on terms that are humane, regulated, and responsive to the people it affects. So, here is what I think the 2026 graduation season actually revealed. It is not a story about tone. It is not a story about which speaker got a messaging right. It is a story about a society that has lost the ability to hold complexity, that has collapsed one of the most consequential technological shifts in human history into a binary of cheering and booing, and in doing so has made itself less capable of navigating what's actually happening. The graduates who booed are not wrong. The world they're entering is genuinely more difficult than the one their parents entered, and AI is one real factor in that difficulty. Wozniak, who was cheered, is not wrong, either. Human intelligence is the substrate on which all of this was built, and centering that is both accurate and necessary. These two positions are not contradictory. They are complementary, and the ability to hold them both simultaneously, to be concerned about AI's impact and to recognize its potential, is the actual failure of discourse. The question, again, is not pro-AI or anti-AI. The question is what kind of people do we want to be? People who reach for the comfort of certainty, who pick a tribe and settle in, or people who can tolerate the discomfort of nuance long enough to actually understand what is happening, and direct our energy where it matters. A college in Arizona replaced the person who says your name at your own graduation with a machine. The machine couldn't do it. A human had to be brought back in to finish. That is not an argument against AI. It is an argument for asking a different question. Not can a machine do this, but should a machine do this? Who made the decision? Who benefited from it? And did anyone ask the students, the people whose names were actually being said, what they wanted? Those questions apply to the scale of a graduation ceremony, they apply to the scale of an economy, and the graduates who learn to ask them precisely, insistently, without retreating into the comfort of a binary, will be the ones who shape what comes next. But Glendale Community College is not the only institution that put AI in charge and watched it fail. Companies across industries have done the same thing, handed critical operations to AI, removed the humans, and discovered that the results were not what the brochure promised. I made a video about what actually happens where organizations go all in on AI replacement instead of augmentation and why the most expensive mistake in AI right now has nothing to do with the technology. That's the one that I would watch next. Thanks so much for watching this one. Subscribe and I'll see you all on the next one.