Fear Is Not Advice
In September 2025, I wrote that education was creating an unemployable generation. Nine months later, nothing has changed.
In May 2026, former Google CEO Eric Schmidt told University of Arizona graduates that AI will reshape every profession, every classroom, every hospital. The crowd booed. He paused, acknowledged their fears were “rational,” and urged them to shape the future of AI rather than reject it.
Weeks earlier, real estate executive Gloria Caulfield was booed at the University of Central Florida for calling AI “the next Industrial Revolution.”
Two stages. Two speakers booked to deliver hope. Two audiences that responded with discontent before the speeches ended.
The reaction has a reason. Across Hong Kong, full-time job vacancies suitable for university graduates fell from 80,000 in 2022 to 31,000 in 2025. Administration roles dropped nearly 90%. IT and programming fell 80%. In Singapore, entry-level ICT job postings have contracted sharply as roles get restructured around AI, with the information technology sector outlook dropping to 15%, reflecting widespread concerns about artificial intelligence replacing entry-level roles. Global consulting surveys point to the same trajectory across multiple markets.
For a graduating student, the math has become brutal in its symmetry. The date of graduation is, increasingly, the date the job market closes. Four years of tuition, exams, and projects, and the position they trained for has been restructured out of existence by the time they cross the stage.
The boos are not anti-technology. They are the right reaction to the wrong message.
Fear Is Not Advice
The default message from speakers, governments, and employers has hardened into one tone: adapt or be left behind.
It is meant as motivation. It lands as a threat.
Telling a 22-year-old to “shape” the future while showing them a labor market that has already locked them out is not a call to action. It is a confession that the people in charge ran out of better ideas.
Fear is not advice. It does not tell a graduate what to do next to continue the journey. It does not name the work that needs doing, the skill to build, or the role to aim for. It only tells them the ground is moving.
What Should We Celebrate For Graduation?
Teachers teach. Students learn. A commencement marks the work both sides did together.
When a graduating class boos, that arrangement has broken. The students are signaling that the journey did not deliver what was promised.
It is worth asking who failed first.
Education Has Been Falling Behind for Decades
This is not a new failure. It is the latest version of an old one.
Education did not absorb every innovation since the beginning of digital evolution. It did not pick up digital practices, e-commerce, or the economics built within the digital ecosystem. It did not adapt to data analytics. Each wave produced the same response: defensive policies, bolt-on certifications, syllabi updated without rebuilding the pedagogy underneath.
What got built instead is a credentialing machine. Degrees, certificates, accreditation. The bureaucracy of education grew faster than the practice of it. Knowledge building has not been the design center for a long time. Compliance with credential frameworks has.
AI did not cause this gap. It exposed it.
The Pedagogical System Is Collapsing
The question, plainly: is the pedagogical system collapsing?
Our intention to build knowledge has not kept pace with the evolution of intellectual advancement around it. AI sharpens the gap. The technology is now at a stage where knowledge transfer can be transformative through self-improvement methods. A student with a model and a serious question can iterate, test, and refine understanding at a pace no traditional classroom matches. The mechanics of transfer, the part teaching has historically owned, is being done elsewhere.
Here is the symmetry the conversation keeps avoiding. Students face a job market judging whether they can produce value AI cannot. Teachers face the same judgment against the same standard. But the industry rarely warns teachers to adapt or be left behind.
The protections the teaching profession has relied on, accreditation, tenure, institutional inertia, will hold for a while. They will not hold indefinitely.
“Learn AI” as Tool Training Is Naive
Less than a year ago, universities were treating AI use as plagiarism.
In July 2025, Singapore Nanyang Technological University upheld a zero mark for a student after a panel found 14 instances of false citations or data in her essay. NTU said the errors were commonly associated with generative AI tools, which were explicitly prohibited for the course. The student was penalized.
Now the same bureaucratic system is warning students that they are not using AI enough.
Students penalized last year for touching AI are being told at graduation this year that the future belongs to those who embrace it. The signal flipped from “do not use this” to “you are behind if you do not use this” in less than twelve months, with no curriculum built to bridge the two positions.
This is the context for “learn AI” as advice. What gets delivered is tool training. How to prompt. How to generate. How to operate whichever model launched this quarter. By the time a course is built, the model has changed.
Operating an AI tool is not difficult. The interfaces are designed to be frictionless. What is difficult is knowing what to ask, recognizing what is wrong, and verifying the output before using it. The NTU student did not get a zero because she used a tool. She got a zero because the tool produced citations that did not exist, and she submitted them without checking. The tool was fluent. The student trusted the fluency. The work of verifying was skipped. The question is, did we train the student how to scrutinize the data?
That work cannot be taught by another tool tutorial. It comes from knowing the domain well enough to spot when the output is wrong, and from the discipline of checking before submitting. Tool training delivers neither.
Training a generation to operate tools without building this discipline is not preparing them for the future. It is producing operators. Operators are the easiest layer of any workforce to automate next.
AI Is a Byproduct of Knowledge
AI is not the source of knowledge. It is a compression of it. Every model on the market was trained on text humans wrote, code humans debugged, papers humans peer-reviewed, decisions humans documented. The intelligence is borrowed. It was produced upstream, by people thinking, testing, failing, and recording what they learned.
Without that upstream activity, there is no model. AI is a byproduct of human knowledge. Treating it as a replacement for the process that creates knowledge is a category error.
The question that follows is uncomfortable. Who is responsible for producing the knowledge in the first place? That work has always belonged to teaching. Not the transmission of content, which AI now handles at scale, but the building of reasoning, the challenging of assumptions, the discipline of verifying before trusting. This is the work the labor market is openly demanding and quietly unable to find.
The institutions responsible for producing knowledge hold the most strategic position in the entire AI economy. Not the platform companies. Not the tool vendors. The places where reasoning and evidence are still built one student at a time. Yet the response to AI in education has been consistently defensive. Detection software. Use policies. Tool licenses. The conversation is still about defending the old format, not building the new one.
Students notice. They are using AI more fluently than their teachers. Then they are told, at graduation, that the future depends on their willingness to adapt.
The boos are the response to that contradiction.
A graduation ceremony is the moment we celebrate actual intelligence. The kind humanity has used to live, build, and grow for as long as there has been a humanity. The kind that four years of study were meant to develop. The kind that walks across the stage in a gown.
Two speakers stood in front of that ceremony and urged graduates to embrace artificial replication of themselves. Apple Co-Founder Steve Wozniak did the opposite. Wozniak told the Grand Valley State University class of 2026 that they already had AI - Actual intelligence. The crowd applauded.
Actual intelligence is what humanity has always used to live and grow. Artificial intelligence is what we built to assist that growth. One is the source. The other is the tool. A graduation that confuses the two is no longer celebrating the right thing.
The work ahead is to train humans to use AI for growing value. Not to blame them for not running fast enough. Not to hand their minds over to the machine. The intelligence is already in the room. The question is whether the next generation of teaching can recognize it and build on it.
Until that happens, the boos at the next commencement are already booked.
