by Students came back to Columbia College this September with printed books and pens. In Literature Humanities, phones stayed in bags. Laptops stayed home or shut inside backpacks. Some of the paperbacks were already creased. You could hear pages turn. For the length of each seminar, the only networked machine in the room was the human being across the table. The same rule now governs every section of Contemporary Civilization, the sophomore course that asks twenty-odd students at a time to argue about Plato and Du Bois face to face.
The news arrived through the campus paper. On September 10 the Columbia Daily Spectator reported that Columbia College had prohibited electronic devices in all sections of both courses for the 2026-27 academic year. There was no press release. The college communicated the policy to instructors over the summer and supplied recommended syllabus language, and a spokesperson described the rule as provisional, a one-year experiment. Music Humanities has enforced a similar ban since 2025, which makes this fall an expansion of a pilot already running.
The headline version of this story says Columbia University banned technology. The record says something narrower. One undergraduate college, two Core Curriculum courses, one provisional year, with exceptions for instructors who have a specific pedagogical use and for students whose accommodations run through disability services. Dennis Tenen co-chaired the faculty working group behind the policy. The professor of English and comparative literature told the Spectator that instructors still have the freedom to use technology in smart ways, citing the example of a group annotating a shared text on screen. The ban is a fence around two rooms.
Richard John gave the Spectator the scene that forced the issue. The historian teaches Contemporary Civilization and served on the working group. A student would look at an electronic device during discussion, punch in some letters, and read back an answer generated by an AI tool. John called the practice “very insulting to the instructor,” and the working group heard “quite a lot” about it. The words arrived on time; the person was somewhere else. Then he asked the question the policy answers: “Is that the kind of people we want to be when we grow up? Is that the kind of people that we want to interact with?”
Columbia built the country’s first college general education program. It began as a single course, Contemporary Civilization, launched in 1919 to confront what its designers called the insistent problems of the present. That course grew out of a War Issues class built for the Student Army Training Corps during the First World War. The Core began as an argument about what a student needs to know before being sent into the world. Literature Humanities joined in 1937. Art Humanities and Music Humanities followed in 1947. Frontiers of Science arrived in 2004. Through every revision, the method held: a small seminar, a difficult book, a conversation carried by the people in the room. Generations of students learned to argue in those rooms before any machine could argue back.
Clémence Boulouque co-chaired the working group with Tenen. An associate professor of Jewish and Israel studies, she framed the ban as an opportunity to make the Core a place of greater intentionality. She conceded to the Spectator that it might feel like a moment of “eat your veggies,” and she stood on the principle: “this is also what we think the humanities should set the standard for.” Chloë Kitzinger, an associate professor of Slavic languages who teaches Literature Humanities, had spent years asking students to keep laptops away on her own authority. She welcomed a coursewide rule because it levels the playing field. A private request from one professor is a favor students can weigh.
Nicole Wallack, who directs the Undergraduate Writing Program, warned that device-free classes are not automatically AI-resistant classes, and that no single set of principles fits every section. Boulouque herself doubts Frontiers of Science would accept the rule, and she noted that Art Humanities students legitimately zoom in on images of works they study. The ban’s architects know its borders.
The most quoted study in the debate is a 2017 paper in the Journal of the Association for Consumer Research. Adrian Ward and three colleagues reported that the mere presence of a person’s own smartphone reduced available cognitive capacity, even when the phone sat face down and untouched. The finding was named brain drain, and it traveled far. In 2022 a pre-registered replication published in Acta Psychologica reran Ward’s second experiment with tighter controls and found no effect. The phone on the desk may be less poisonous than the famous result claimed.
In 2013, Faria Sana, Tina Weston, and Nicholas Cepeda reported in Computers and Education on students who multitasked on laptops during lectures. The multitaskers scored lower on comprehension tests. Students seated in view of a multitasking peer scored lower too. Your screen taxes me. That secondhand finding is the one a seminar lives or dies by. The second result is the 2014 Psychological Science paper by Pam Mueller and Daniel Oppenheimer. Students taking longhand notes outperformed laptop note takers on conceptual questions, with the advantage widening after a week. A 2019 replication by Kayla Morehead and colleagues found the longhand advantage smaller and narrower than first reported. The effect survived, smaller.
Distraction is measurable, and it spreads across a room. The mystical part, the phone that drains you by existing, did not replicate. Credit the working group for this: it rested its case on witnessed behavior. Students were prompting chatbots mid-discussion and reciting the returns. A policy that answers a documented classroom practice needs no help from a brain in a scanner. The seminar is a commons of attention. One glowing rectangle withdraws from every account at the table, and a rule that removes the rectangle protects people who never touched it.
Students with accommodations from disability services keep their devices. Instructors keep the right to open a screen when a lesson calls for it. A rule with no exceptions would be a gesture. This one keeps its exceptions where everyone can see them. A lecture hall cannot police itself. In a room of twenty, a screen is a choice visible to everyone, and visibility does half the enforcing.
I hold a degree from Columbia’s School of the Arts, class of 1991. I have earned my living for decades writing about computers, and my books sit on the same shelves as the machines they explain. Apple shipped the first PowerBook on October 21, 1991, weeks after I finished. My Columbia seminars ran on paper because paper was what the world had built. Nobody in those rooms chose the analog life. We inherited it, the way every class before us had. There was the book, and there was one another. I spent those years at workshop tables where the pages in front of us were the only evidence admitted, and where a sentence had to survive being read aloud by its author in front of its victims. That training set the terms of my trade. I have never confused the instrument that delivers a text with the mind that must answer for it.
I support this ban. My income depends on the machines I write about, and I do not find the two hard to reconcile. Tools belong to rooms. A networked screen is an instrument for research, composition, and reference. It belongs to the library, the studio, and the office. The seminar table runs on a different instrument, one we spent five centuries perfecting: the printed book, marked up, carried in, argued over aloud. Writing about machines teaches you one thing early. They are built by people paid to hold your gaze. A student who asks a chatbot for a sentence during a discussion of Homer has left the conversation while keeping a chair in it. The machine will be waiting after class. I will be waiting with it, professionally. For two hours, the room gets the people who showed up.
In May 2026 the law school at the University of California, Berkeley, made nonuse of AI the default for any work submitted for credit. A month later, the dean of the University of Texas law school urged his faculty to keep students off their screens during class. July brought the University of Chicago Law School’s no-device policy across all nine of its core first-year courses, coupled with oral examinations. The move belongs to a strategy the school calls Rethinking Legal Education in the AI Era. Dean Adam Chilton told Inside Higher Ed the rules “force students to think for themselves, be able to answer a question on the fly.” Columbia’s September policy arrived at the crest of a wave already moving through professional schools.
Twenty years ago the Christian Science Monitor reported a small rebellion. Two Harvard Law School professors had independently banned laptops from their classrooms. The University of Michigan’s law school had blocked wireless internet in class, joining UCLA and the University of Virginia. Linda Stone, a former Microsoft executive, gave the condition its name in that era: continuous partial attention. The diagnosis is old enough to vote. What changed is what waits on the other side of the screen. In 2006 the laptop hid email. In 2026 it hides a ghostwriter.
The Ivies were stirring before Columbia moved. In April 2026 Yale’s Committee on Trust in Higher Education published a year-long inquiry co-chaired by the sociologist Julia Adams and the historian Beverly Gage. The committee spent that year listening, through public comment pages, stakeholder sessions, and a traveling speaker series. Its case rested on a collapse in public confidence. Gallup’s tracking ran from 57 percent of Americans in 2015 down to 36 percent in 2024, a figure that climbed to 42 percent in 2025 before slipping back to 38 percent this year. Among its twenty recommendations sat one blunt sentence: a device-free policy, “no phones, laptops, or tablets,” as “the default in classroom settings,” with faculty free to depart for stated reasons. The committee wrote that the technologies of the century had transformed teaching in many positive ways, and that constant distraction was not one of them. President Maurie McInnis accepted some recommendations in full and referred others to committees. Some she left unaddressed. Where Yale’s device default lands remains an open question this fall. Harvard’s Bok Center for Teaching and Learning has advised faculty since at least 2024 to limit or prohibit internet-enabled devices, citing the research on learning losses, while leaving the mandate unwritten. Princeton governs AI through disclosure rules and instructor discretion. Penn, Dartmouth, and Brown leave the question to individual professors, and student newspapers at places like Washington University report device bans spreading through arts and sciences syllabi one classroom at a time.
A device ban costs nothing and polls like a tax cut. Twenty-two states and the District of Columbia had bell-to-bell phone bans in their schools by June 2026, according to the Harvard Kennedy School, and a Pew survey released this year found 77 percent of Americans backing classroom phone bans. Administrators at the remaining Ivies will not announce that they are copying Columbia. They will announce task forces. The task forces will discover Yale’s report. Within two academic years most of the Ancient Eight will have a device-free default dressed in local colors. The working word will be default. A default lets every professor opt out while making every student opt in.
I have watched NYU shadow Columbia for forty years. Uptown announces, and downtown answers a season later. This time the shadow will not move. The reason is structural. NYU has written AI into its institutional identity. In November 2022 the university committed a billion dollars over twenty years to its Tandon School of Engineering, with data science named among the priorities. In January 2024 NYU joined Governor Hochul’s Empire AI consortium as a founding member, a public-private computing venture that has since grown past five hundred million dollars. Its Courant Institute and its Center for Data Science anchor its research identity in machine learning. NYU’s Steinhardt school publishes official guidance on the question. It tells faculty that “instead of banning AI, educators should focus on teaching students to use AI as a learning tool.” A university cannot sell fluency in a machine with one hand and clear that machine off its seminar tables with the other.
The irony has a name and an office. Jonathan Haidt, the social psychologist whose 2024 book The Anxious Generation armed the phone-ban movement across American schools, teaches at NYU Stern. The prophet of the device-free classroom works for the university least able to adopt one. Haidt’s argument persuaded legislatures because it addressed children. NYU’s hesitation concerns adults and a brand staked on the promise that its graduates will build the AI economy, and hiding from that economy sells nothing. Both things can be true at once. Only one of them fits on an admissions brochure.
Columbia sits inside Empire AI too. Its IT department now offers the campus ChatGPT Edu, Claude for Education, and Google Gemini. The provost’s office has tied teaching awards to AI course design, according to Bwog’s reporting in May. The law school announced in August that students may use AI as a learning aid, provided nothing submitted for credit was generated by it. On July 1 the university’s twenty-first president took office. Jennifer Mnookin is a scholar of legal evidence, a person who has spent a career asking what counts as proof, and her years leading the University of Wisconsin included major campus initiatives on artificial intelligence. The same institution that fenced the seminar table buys enterprise seats from the machine’s makers. That reads as hypocrisy only if you believe a university is one room. It is many rooms, and Columbia has decided, correctly, that the oldest room it owns gets a different rule.
The policy is provisional. That is the smart part. Surveys of students and faculty came before the recommendation, and the working group’s full report still awaits the committee that governs the Core. The college built itself an exit and a metric at the same time. If the seminars go flat, the rule can die in a year of quiet noncompliance. If the rooms come back to life, the record will show it. I know which outcome I expect. Somewhere in Hamilton Hall this week, a sophomore will look up from a creased paperback, catch a classmate’s eye over an open question, and answer it with a sentence no model wrote. The glass stays off the table. The conversation stays in the room. That was the whole idea in 1919, and it is the whole idea now.
Leave a Reply