There is a crisis building inside higher education, and it has nothing to do with budget cuts, enrollment cliffs, or campus politics.
It’s about what happens when an entire generation of students learns to think with — and eventually through — a machine, before they’ve learned to think for themselves.
I work across both higher education and manufacturing. I see AI from both sides of the pipeline: the institutions preparing the workforce, and the companies absorbing it. And what I’m seeing should alarm anyone who cares about the quality of human judgment in the economy we’re building.
The rise of algorithmic complacency
There is a term that doesn’t get enough airtime: algorithmic complacency. It describes the gradual, almost invisible erosion of critical thinking that happens when people stop questioning the output of an automated system.
Pilots experience it with autopilot. Radiologists experience it with diagnostic AI. And now an entire generation of students is experiencing it with large language models — before they’ve even entered the workforce.
The pattern is consistent and predictable. A student uses AI to draft a paper. The output is coherent, well-structured, and superficially impressive. The student submits it with minor edits. The grade comes back acceptable. The lesson learned is not about the subject matter. The lesson learned is that the machine is good enough.
Repeat this a hundred times across four years of higher education, and you don’t produce a graduate. You produce a prompt operator with a degree — someone who can extract output from a system but cannot evaluate whether that output is true, relevant, original, or dangerous.
This is not a technology problem. This is a human development problem enabled by technology.
Higher education’s governance vacuum
Here is the uncomfortable reality: higher education — the institution society entrusts with preparing critical thinkers for the workforce — does not have a generally approved ethical framework for AI use.
Not at the institutional level. Not at the accreditation level. Not at any level that would give a faculty member, a student, or an employer confidence that the rules are clear, consistent, and enforceable.
Some institutions have written AI use policies. Most of those policies are reactive, vague, and unenforceable. “Use AI responsibly.” “Cite AI-generated content.” “Don’t submit work that isn’t yours.” These aren’t governance. These are suggestions dressed up in academic language.
The problem isn’t that institutions don’t care. It’s that the technology moved faster than the governance conversation. Faculty are expected to set boundaries they don’t fully understand, for tools they didn’t choose, in a regulatory environment that doesn’t yet exist.
And the students — who are, by definition, still forming their intellectual habits — are left to self-regulate their use of the most powerful cognitive shortcut ever invented.
We would not ask a first-year medical student to self-regulate their use of surgical tools. But we are asking 18-year-olds to self-regulate their use of AI systems that can produce convincing arguments on any topic, in any discipline, at any level of apparent sophistication.
What this means for the workforce
I’ve sat in manufacturing boardrooms where the conversation is about hiring engineers who can think independently, solve novel problems, and exercise judgment under ambiguity. These are the skills that justify a human salary in a world where automation handles the routine.
Now project forward five years. The graduates entering those boardrooms — and those factory floors, and those IT departments — were trained during the era of uncritical AI reliance. Not all of them, of course. But enough of them to shift the distribution.
The risk isn’t that AI replaces workers. That narrative is overplayed. The risk is that AI degrades the quality of the workers it was supposed to augment. Not because the technology is flawed, but because we deployed it into the learning environment without guardrails, without governance, and without honest conversation about what we were trading away.
Differentiation value — the ability to see what others don’t, to synthesize across domains, to challenge assumptions and propose alternatives — is the one asset that no AI can replicate. It’s also the asset most directly threatened by algorithmic complacency.
If a student has spent four years accepting AI output without questioning it, they will spend the next forty years doing the same thing. And the companies that hire them will feel the cost — in missed insights, in unchallenged assumptions, in decisions that looked smart because the machine said so.
What needs to happen
I don’t have a silver bullet. But I have a perspective shaped by sitting on both sides of this pipeline — the education side and the employer side — across multiple industries and continents.
First, higher education needs AI governance that is specific, enforceable, and tied to learning outcomes. Not “use AI responsibly” — but clear, course-level guidelines that articulate when AI use supports learning and when it substitutes for it. This requires investment in faculty development, not just policy documents.
Second, the conversation needs to shift from detection to formation. The plagiarism-detection arms race is already lost. AI-generated content is indistinguishable from student-written content in most contexts, and the detection tools generate false positives that damage trust. The better question isn’t “did the student use AI?” It’s “did the student develop the cognitive skills this course was designed to build?” That requires rethinking assessment, not just policing tools.
Third, employers need to enter this conversation. The companies that will hire these graduates have a legitimate stake in how they were trained. Industry advisory boards, employer-informed curriculum reviews, and honest feedback loops between hiring managers and academic programs aren’t new ideas. But they’ve never been more urgent.
Fourth, AI literacy must become a foundational skill — not a technical elective. Every student, in every discipline, should graduate understanding how AI systems work, where they fail, and why uncritical reliance is a professional liability. Not because every student will build AI. Because every student will use it.
The deeper question
The question underneath all of this isn’t about technology. It’s about what we value.
If we value efficiency above all else, algorithmic complacency is a feature, not a bug. Let the machine think. Let the human operate the machine. Optimize for speed.
But if we value independent judgment, original thought, and the ability to challenge consensus — the very things that drive innovation, that create new markets, that solve problems nobody knew existed — then we have to be honest about what we’re trading away when we hand a 19-year-old an AI assistant and call it progress.
Higher education is the last institution that has the mandate, the time, and the authority to develop these capacities in young people before the market gets hold of them. If it doesn’t rise to this moment, nobody else will.
And the workforce — and every company that depends on it — will pay the price.
Gábor Szentiványi is a Transformational CIO and Certified AI Consultant with 25+ years of enterprise IT leadership across manufacturing, higher education, consumer goods, and pharma. He works at the intersection of AI strategy, IT governance, and organizational change across three continents.