High On AI

Ethics in Columbia’s AI Minor 

By Isabelle Oh 

Illustration by Isabelle Oh

Many of my classes have been transitioning away from essays, projects, and take-home tests, opting instead for the infamous blue-book final-exam format, handwritten notes in black pen, all in an effort to curb the use of artificial intelligence in class. 

But I see the technology used, regardless of our professors’ wishes, all over campus. In addition to the online shopping or spectacular fails at speedrunning Minesweeper, when I look down from my perch in the back row of a lecture hall, I see text tumbling onto computer screens, firing off AI-generated emails or automatically taking notes on lectures. 

So far, it seems that professors are only able to mitigate AI use in their classes by making sacrifices to the quality of assignments and tasks or erecting flimsy walls in paragraph-long provisions buried in a syllabus. Many of my professors acknowledge that, if not for AI, a midterm or final would have been a more comprehensive and appropriate essay assignment. Instead, exams lean toward memorization rather than contemplative analysis, because at least there’s something in our brains—even if it is just regurgitated facts. Other professors structure classes and assignments so that homework makes up less of the final grade or emphasize that using AI for an assignment will make studying for the final more difficult. Even some computer science professors have rejected AI in the classroom. Notoriously, Professor Jae Woo Lee, who teaches Advanced Programming, enforces a strict no-AI policy in his class. He argues that even if AI is able to write code that produces the desired effect, it deprives students of the “cognitive struggle” of getting to the solution. Instead, as he told The Eye, AI is really only useful to those who have already developed a full understanding of coding. Despite these policies and philosophies, numerous students were found by Professor Lee to use AI in Advanced Programming (running to their fellow human beings on Reddit to help predict the repercussions). Faculty seem to want to control AI use in their classrooms but have not yet found effective ways to do so. My observations from the back of the lecture hall tell me that professors are still playing catch-up.

Not all students embrace AI use. In fact, the University’s incorporation of AI into its operations has not been received without resistance and debate. This past May, student groups protested the University’s use of an AI program called Tassel to read graduates’ names on Class Days and acquisition of a Claude Pro subscription, or “Claude for Education,” which the announcement states “helps you think, write, code, and solve problems.” Student Workers of Columbia, UAW Local 2710, the union representing student workers, has included a provision on AI use in their ongoing negotiations for their upcoming contract with the University. As of July 9, SWC has demanded that AI not replace student workers, not be used to justify decreased hours or pay, and that student workers not be compelled to provide an AI with their material or be forced to use it. Students and teachers alike are pushing back and demanding the University respond with some level of caution and containment. 

So I was confused to learn that in November 2025, Columbia Engineering announced they were implementing an Artificial Intelligence Minor for non-computer science majors in the engineering school. Students must take various math, computer science, and engineering courses, in addition to an ethics requirement.

I was most struck by this peculiar addendum of an ethics requirement to a minor that consists of more p-sets than discussion sections. Most of the classes for the ethics requirement surprisingly fall under the Computer Science Department. Ethics, if we are sticking to the classic divisions of academic disciplines, is usually under the realm of philosophy. One class falls under the Industrial Engineering and Operations Research Department, and another is part of the Psychology Department. The term “ethics” appears in the title of two other classes for this requirement: “Ethical and Responsible AI” and “Large Language Models: Foundations and Ethics,” both part of the Computer Science Department. Should an “ethics requirement” not in some way involve an ethics researcher? The involvement of ethics seemed like a layer of plastic wrap, thin and penetrable with a poke. 

Education, as it contends with AI, is left in a sort of double bind. The humanities, where the study of ethics traditionally resides, opposes AI and refuses to use it in the classroom as it truncates critical thought. STEM or STEM-adjacent departments, such as computer science, data science, and economics, are structurally more geared towards incorporating AI into their research methods and the classroom. The disciplines that can likely teach the most about what ethics in artificial intelligence looks like refuse to use AI and thus participate in teaching AI’s ethics, while disciplines that rarely contemplate the consequences of number crunching are where AI is used the most. A rift, then, splinters the University, disparate halves talking at, but not with each other. 

The syllabi for the classes in the ethics requirement focus primarily on data privacy, limitations of the technology, and appropriate technology use, among other skills. “LLMs: Foundations and Ethics” incorporates readings about how large language models are built, how they work, and their limitations, as well as biases in training data, AI’s effects on labor, how to better prompt LLMs to produce more objective outcomes, and plagiarism in LLMs. The 2024 syllabus for “Policy for Privacy Technologies” addresses issues of encryption, anonymization, use of synthetic data (made-up data intended to mimic the real world), and data flows. Most of the descriptions for these “ethics” classes include units on containing and manipulating the technology, how the data on which these technologies are trained influences the kinds of outputs, and how those outputs betray the biases of the data it was trained on. 

Privacy, bias, and plagiarism are all problems that involve ethics, or attempt to discern between right and wrong. This approach to ethics may be too limiting for a technology that we have begun to accept as inevitable. It does nothing to address the enforced, chaotic, tumbling urgency of assimilation, squeezing our to-do lists and junk email folders and discussion posts and papers and books and creative works through a funnel of hollowing acceleration. There is a constant whispering: “Do more, all the time, without thinking.” 

The foundation of this ethics requirement lies on an assumption of the inevitability of AI. But evidently, we haven’t completely figured out how it will work alongside, against, or perhaps in favor of education and teaching. 

To the world, the creation of this minor presents the University as supposedly ahead of the curve. It puts forward an acceptance of the current form of artificial intelligence as the next “logical” step in technological development. AI is often presented as an inevitable fact of life, as something we must learn to cope with or else risk “falling behind.” The University is pushing an agenda that it cannot keep up with, at least in terms of maintaining a quality of education that actually encourages students to be invested in their learning. Factions of the University remain stagnant in a series of practices that the rest of the world seems to want to make obsolete, while other departments are feeding off of AI like vampires. This growing chasm within the University reflects a more foundational problem: our current study of AI doesn’t consider the market forces and motivations for its incorporation into academia and industry. It is a consideration the University must make before it can even think about jumping ahead to the creation of an AI minor. 

The larger problem with trying to moralize AI or study the ethical potential of AI is that its fate and direction are determined not by individuals who use it but by the multibillion-dollar companies developing these technologies. OpenAI was founded in part as a competitor to Google’s AI development. Anthropic can make all kinds of claims about their philosophers supposedly giving Claude a set of ethics to empathize with users, but it’s still a multibillion-dollar company that uses Amazon’s and Google’s computing power. Google owns 14% of Anthropic. Microsoft has invested billions of dollars into OpenAI. AI is becoming increasingly ubiquitous, impenetrable, and capacious, touted as the hero or destroyer of worlds. If we don’t keep up, we’ve somehow lost a war. 

Academics, try as they might to turn away from AI’s infiltration into research and teaching, will likely struggle to do so within the flailing university ecosystem—amidst censorship, lack of public funding, and motivations to pursue topics based on their fundability. They might instead be motivated to embrace the technology rather than challenge it. Nvidia’s CEO, Jensen Huang, recently donated $75 million toward creating an art school at Vanderbilt’s new San Francisco location. Ironically, the man whose company is slowly eating away at artists’ livelihoods is funding the schools that should turn out more artists. But the donation represents a dependency on private money for academic research. When it’s AI companies that are providing those funds, academics will inevitably be motivated to include AI in their course design and research methods. Similarly, Columbia’s Office of Teaching, Learning, and Innovation has awarded grants to faculty who incorporate AI into their course designs. For example, of the seven courses given the Spring 2023 Innovative Course Module Design Award, four went to courses with “AI” or “Artificial Intelligence” in the title, compared to none in 2021. Academics can critique artificial intelligence all they want, but the flow of money suppresses and discourages deviation from the prescribed norm. This is not to say that there is no one doing the important work of zooming out from the race of AI development and assimilation and assessing the reasons behind this insane push, but the flows of funding and job security discourage criticism. 

What, then, is education for, if the University that claims to prepare its students to be critical of the world around them falls for the same tricks, accepting this inevitability of AI, finding problems with AI only within the parameters the industry provides? Why debate the ethics of privacy or plagiarism in artificial intelligence when we can and should be thinking about the logic and reasons behind AI’s encroachment into our lives? 

The creation of this minor signals an acceptance of AI into the formal fabric of University education. Not only is a large portion of the University not challenging students’ use of AI, but it also reveals the University’s ongoing trajectory toward an uncritical existence, a willingness to forgo criticism of the institutions that prop up the university industrial complex in favor of playing directly into their hands.