Computing has spent seventy years as the disruptor. Travel agents, print media, retail, taxis, photography — one industry after another watched software dissolve its assumptions while computer scientists explained, with varying degrees of sympathy, that this is what progress looks like. There was always an unspoken asterisk: it won't happen to us. The people who build the wave don't get hit by it.
That asterisk just expired. The strangest fact about this technological moment is that computer science is now the most disrupted discipline on campus — disrupted by its own creation, more thoroughly than it ever disrupted anyone else. Having spent a long time around this field, I find the situation equal parts vertiginous and clarifying, and I want to think through it honestly — including the uncomfortable parts.
Two broken pillars
Universities everywhere are wrestling with generative AI, but for most disciplines the damage is confined to one pillar: homework is obsolete. The take-home essay, the problem set, the lab write-up — all now completable by a chatbot, which means they no longer certify learning, which means the entire assessment architecture built on them is running on inertia. That's serious, and it's universal.
CS faces that plus something no other discipline faces: a crisis of the curriculum's object. The essay-disrupted history department still believes historical thinking matters. But the fundamental skill CS departments teach — coding, the thing we spent a decade telling everyone to learn, the centerpiece of "Coding for All" — is precisely what AI agents now do fluently. The discipline must simultaneously redesign how it teaches and confront whether what it teaches remains a viable human skill. No other field on campus got hit at both levels at once.
And the students have noticed. Doctoral students — people who bet five or more years of their lives on this field — are asking out loud whether they chose wrong, watching research funding wobble, watching the entry-level job market flatten, reading headlines about CS graduates seeking shift work. Their anxiety compounds arithmetically: falling undergraduate enrollment means fewer teaching assistantships means thinner funding for graduate study means fewer academic careers. The whole academic food chain feeds on undergraduate demand, and undergraduate demand feeds on the perception of jobs.
The case for calm — and its limit
Before declaring the end of the discipline, honesty requires the historical detour, because we have been here before — almost exactly. In the early 2000s, computing endured a perfect storm: the dot-com crash, the telecom collapse, an offshoring panic ("all the jobs are going overseas"), and a research-funding squeeze. Enrollments in North American CS programs fell off a cliff. Serious people wrote serious obituaries. The field's institutions convened task forces to study whether the profession had a future — and the considered answer was yes, an optimism the next twenty years vindicated so completely that the industry's giants now carry a market capitalization around twenty trillion dollars. The offshoring scare, in hindsight, was a mispriced panic about a real but manageable shift.
So the cyclical reading is available: panics pass, enrollments recover, the field reinvents and re-booms. Anyone forecasting doom for CS must explain why this time differs from the time doom was forecast and spectacularly failed to arrive.
Here's my attempt at that explanation — because I do think this time is structurally different in one specific way. The 2000s crisis was external: bad markets, scary geopolitics, funding weather. The object of study was untouched; when the weather improved, the discipline stood exactly where it had been, intact and waiting. Today's crisis is internal: the discipline's own artifact now performs the discipline's foundational skill. No market recovery un-invents coding agents. The weather isn't the problem; the ground moved.
What CS is when coding is cheap
But notice what that framing implies — and here the clarifying part begins. If the crisis is that coding got automated, then the crisis falls on the portion of CS that was actually about coding. And the field's own elders have insisted for decades that this portion was always smaller than outsiders assumed. Computer science was never the study of typing programs; it's the study of computation — problem decomposition, abstraction, algorithmic thinking, systems design, the limits of the computable. Coding was the interface to those ideas, and for a long era it was also the economically precious skill, so curriculum and marketing drifted toward it. "Learn to code" was always a compression of "learn to think computationally," monetized.
The automation of the interface forces the discipline back onto its actual substance — and that's a reckoning, not an execution. A curriculum organized around syntax mechanics and routine implementation is genuinely obsolete. A curriculum organized around decomposition, systems reasoning, verification, and the evaluation of machine-produced artifacts is more necessary than it was, because someone must direct and audit the agents. The uncomfortable transition is that the first curriculum is cheap to teach and assess at scale, and the second is expensive — it looks like studios, projects, oral defenses, critique — which is why institutions will resist it until enrollment economics force the issue.
As for assessment: the obsolescence of homework strikes me as overdue honesty. Homework certified artifact production, and artifact production is exactly what automation commoditizes — in the classroom today, in the workplace tomorrow. Whatever replaces it — explain your system aloud, defend a design, find the flaw in this generated code — will measure understanding rather than output. CS, having caused the crisis, could lead that redesign and export it to every other department on campus. There would be some justice in that.
The discipline that disrupted everything finally disrupted itself, and the experience is instructive: it turns out disruption feels less like death and more like an involuntary audit — everything incidental stripped away, everything essential suddenly exposed and priced. Computing's essence — thinking clearly about computation and systems — passed the audit. The comfortable parts didn't. We should say so plainly, to ourselves and to the twenty-year-old deciding this fall whether the field still deserves five years of their life. My answer to that student: the discipline is not dying. Its costume is.
