AI Is Why Computer Science Belongs in the Liberal Arts
The hype surrounding AI often makes it sound like this is the moment when nobody will ever need to truly learn anything again. And this is particularly true when it comes to computer science and especially computer programming. Why bother understanding syntax when a chatbot can fix your missing semicolon? Why think when you can just ask a model to think for you?
In this way, AI adds a new twist to the old “Is college obsolete?” debate. Universities have been circling a related question for decades:
What is college for?
If college is mainly a workforce pipeline, then sure, teach everyone to prompt their way through tasks. But:
If CS graduates should leave with judgment, creativity, and intellectual agency, then computer science belongs next to philosophy and math as a liberal art.
AI Exposed What (Most) CS Education Was Really Doing
When I was an undergrad, there was a recurring debate about where “Computer Science” belonged: Engineering or Arts & Sciences. Schools picked one, usually based on whether they saw CS as professional training (like other engineering fields) or as an academic discipline worth studying for its own sake.
Regardless of what university college the CS department ended up in, though, I think this debate was de-facto settled sometime in the ‘00’s around the point where the industry started paying six-figure starting salaries and enrollments took off as students saw dollar signs on the other side of graduation. (And soon after, you had bootcamps, which were generally lower-quality but at least not hypocritical about what they were doing in terms of, say, interview preparation and resume-building.)
For what CS education has become and where it might be going, the Business Insider interview with Dan Grossman (UW professor and CS educator) is a perfect snapshot of the moment. Grossman points out that AI is already handling the tiny details that used to dominate early CS education like syntax, punctuation, and the “did you put the bracket in the right place” stuff.
He’s right. AI is great at that.
But those details were never the point. Rather, they were the training wheels for learning how to think computationally.
Memo to CS departments everywhere: If your entire educational philosophy collapses because a chatbot can autocomplete a for‑loop, maybe your educational philosophy wasn’t that great to begin with.
Two Competing Models of CS Education for the AI Era
Once you take AI seriously, the question isn’t whether CS classes will change, because they will. The real question is a modern version of the old engineering-vs-arts-and-sciences debate, namely, what kind of education we think CS should be.
1. College as Workforce Pipeline
Teach prompt engineering
Teach basic skills for testing and deploying systems.
Teach whatever the job market is screaming about this quarter
Pretend this is “education”
That might be ok for awhile, but it leaves graduates especially vulnerable to being made obsolete by the very AI they are being trained to operate.
2. College as Liberal‑Arts Formation
Teach reasoning
Teach abstraction
Teach systems thinking
Teach how to challenge assumptions
Teach how to model the world
In this model, AI is a catalyst. It forces us to double down on the parts of thinking that cannot be outsourced.
What I’m afraid will happen is that universities will pretend to do #2 while quietly drifting toward #1 under pressure from industry backers, students, and maybe governments.
Yes, the Semicolons Still Matter
A lot of people in industry and education seem really excited about not having to worry about “where semicolons go” anymore. But I actually think syntax is incredibly relevant even in the age of AI. Learning “where the semicolons go” and “how to organize brackets and parentheses” is part of the cognitive training. It’s not about the semicolon. It’s about:
precision and discipline
attention to detail
understanding structure
developing an internal model of how systems behave
learning to debug your own thinking
AI can fix your syntax. It cannot fix your mental model.
If you’ve never internalized the structure of a program, you won’t know when the AI is hallucinating, overgeneralizing, adding a new layer of mediocre code to paper over the previous layer of mediocre code, or quietly plotting to delete your production database and backups.
I took piano lessons as a kid. I played scales. The semicolons are the scales of software engineering.
Why We Should Finally, Officially Treat CS as a Liberal Art
Take away the hype, buzzwords, and frameworks and Computer Science and Software Engineering are fundamentally about:
logic
abstraction
representation
constraints
tradeoffs
modeling complexity
designing systems
These are the same intellectual muscles exercised by philosophy, linguistics, and mathematics. They’re not “skills” in the narrow sense. They’re ways of seeing the world and operating with in as a productive member of society.
When you learn how a compiler works, you’re not learning it because you’ll write compilers for a living. (Very few people have written compilers for a living in decades.) You’re learning it because it rewires how you think about causality and structure.
When you study algorithms, you’re not memorizing recipes to help with LeetCode interviews. You’re learning how to reason about efficiency and the shape of problems.
When you design a system, you’re not just building software. You’re practicing the art of modeling reality and building useful abstractions.
This is the intellectual core of CS. And especially in the age of AI it looks a lot more like the liberal arts than like job training.
I Started With a Math Degree
Before I later returned to school to get my Ph.D. in Computer science, I earned a liberal‑arts bachelors degree in Mathematics, which included classes in history, philosophy, art, etc. On paper, that said nothing about my ability to do “software engineering”. I can credit my subsequent career success to the fact that my degree taught me how to:
read closely
write clearly
think abstractly
argue rigorously
understand systems of ideas
question assumptions
build mental models
communicate complexity
Depending on how you count, I’ve programmed professionally in 10 different languages, zero of which I actually studied in college. The aforementioned liberal skills turned out to be far more durable than any studying any specific programming language or framework would have been. They gave me leverage. They gave me autonomy. They gave me the ability to navigate a field that reinvents itself every five years.
AI only reinforces this truth, because…
AI Makes Liberal Arts Higher-Leverage
There’s always been a line of thinking that humanities are irrelevant to the modern world, and the AI wave adds a new wrinkle to this argument. There’s a popular fear that AI will make the humanities irrelevant, all knowledge past-present-and-future is encoded in the model, etc. etc. I think the opposite is true.
When AI tools can “do” everything, the only defensible educational mission is teaching people how to think.
AI can autocomplete code and generate text. It can’t autocomplete judgment and generate taste.
The non‑automatable parts of human cognition -- the parts that define leadership, creativity, and originality -- are exactly the parts the liberal arts cultivate.
So What Should Colleges Do?
If we take these changes seriously, CS education needs a philosophical realignment: Treat syntax not as a substitute for thinking, but as a gateway to it. Spend less time chasing whatever framework is fashionable and in industry demand this semester and more time on enduring concepts. Stay away from “learn to code/prompt” in favor of “learn to think computationally,” shift from operating tools to designing systems, and focus less on “how to use AI” and more on “how to reason in a world where AI exists.”
This doesn’t mean abandoning practical skills entirely. It means anchoring them in a deeper intellectual foundation that the next great AI breakthrough can’t automate away.
AI Forces the Issue: Train Workers or Educate Minds?
AI is forcing Computer Science departments to confront a question they’ve danced around for decades:
Are we educating workers, or are we educating minds?
If the answer is “workers,” then by all means, teach AI tools, teach automation, teach whatever the job market demands this quarter.
But if the answer is “minds,” then CS belongs squarely in the liberal‑arts tradition. And the liberal arts belong squarely in CS.
Because the future of software isn’t about typing code. It’s about understanding systems. It’s about modeling complexity. It’s about making choices under uncertainty. It’s about judgment.
And judgment is not something you can outsource to your chatbot.
