We still don’t have a legally certified tool designed to detect AI-generated writing, but Dartmouth College has just rendered a decision that sets a major precedent that is already reverberating throughout academia. The Ivy League university has accepted the resignation of its now-former provost Santiago Schnell, following an investigation into his use of AI to generate academic publications and an opinion piece in The Washington Post. (Ironically, Schnell’s WashPo op-ed was titled “Universities are fighting AI cheating. But there’s a deeper problem.”)
However, the committee investigating the Venezuelan professor of mathematics did not use Pangram, or a tool like it, in their research. Instead, the Dartmouth group based its findings on something distinctly non-technological: Honesty, or the lack thereof. Dartmouth specifically looked for instances in which Schnell used AI, but did not disclose his use of the technology to meaningfully produce and editorially change his writing. As a result, the Advisory Committee’s report found that, “the Provost did not sufficiently disclose his AI use to publishers in contravention of those publications’ policies.”
🙋🏻♂️🤖 MAN AND ROBOTS lands in your inbox every Friday: AI, Hollywood, and the future of work. Get it free →
Like the Watergate political scandal of old, in this case, AI wasn’t the crime, but it was the cover-up of AI use that proved to be an unacceptable breach in the school’s ethics. Disclosure. Theoretically, according to Dartmouth, if Schnell had simply disclosed his use of AI all along, being honest about the work that did or didn’t go into his publications, the outcome might have been different.
Based on this university’s result, the forward-looking academic framework points toward normalizing the use of AI to generate and/or substantially alter the scholar’s written expression, as long as the author discloses that AI was used. In that context, the calculus changes from AI deception (e.g., “I came up with all of this on my own”) to pure informational transaction that may or may not reflect the author’s true expertise in a particular domain.
Disclosing extensive AI-generated writing effectively gives the readers a choice: accept what they openly know is machine-produced, even if partially, as a unique work from a particular human’s POV, or dismiss the writing as statistically generalized AI chatbot outputs that have been lightly tailored by a human opinion.
Right now, there are probably hundreds of college professors around the world secretly worried about finding themselves in Schnell’s position. They’ve found out what just happened to Schnell and are nervously considering how, when, or if they can come forward and disclose their own AI-generated papers without finding themselves jobless.
When contrasting this kind of writing versus the unvarnished, imperfect writing coming directly from a human scholar who doesn’t use AI-generated text, the importance of AI disclosure functions more as an information hygiene nutrition label rather than an ecclesiastical vow of intellectual purity.
Nevertheless, because no AI writing detection software is legally sanctioned (yet), Dartmouth’s honesty-first approach to the Schnell situation, first reported by Semaforback in August, is elegant, effective, and a potentially powerful panacea for other universities to follow.
As students and professors alike increasingly rely on AI to complete assignments, publish academic studies, and even grade the work of pupils, much of the academy has fallen into a state of ad-hoc solutions. Educational responses to AI use appear to range from ignoring it, to scattershot policing it, to even aggressively promoting it, whether or not higher scores lead to real student comprehension (and, ultimately, more competent graduates).
Dartmouth’s approach isn’t perfect, but it’s a start in the right direction. If well-regarded professors like Schnell are writing op-eds in top-tier publications about the pitfalls of AI writing, while themselves quietly using AI to write, who can a student trust to guide them? I think this is the underlying spirit of the Dartmouth committee’s statement in its report.
“Dartmouth has earned and must maintain a reputation for integrity and quality that goes beyond compliance with laws, regulations, and contractual obligations…Even the appearance of misconduct or impropriety can cause severe damage to Dartmouth’s reputation…faculty, staff, and students—are responsible for maintaining a culture of integrity, honesty, and respect in teaching, learning, scholarship, and creative work,” reads the report from Dartmouth’s Advisory Committee, surfaced by The Dartmouth. “The Committee felt that the Provost’s incomplete (or missing) disclosures about his AI use, and public statements characterizing his AI use, are not consistent with “a culture of integrity, honesty, and respect in … scholarship.”
Right now, there are probably hundreds of college professors around the world secretly worried about finding themselves in Schnell’s position. They’ve found out what just happened to Schnell and are nervously considering how, when, or if they can come forward and disclose their own AI-generated papers without finding themselves jobless. Others will just stop using AI to write at all. Still others will more cleverly execute their AI-generated works while hoping for a gradual cultural paradigm shift that blurs the distinction between AI-generated and human-written text, at which point AI becomes broadly institutionalized, and no one cares anymore.
But in real-time, the Schnell episode is meaningful and historic. No, this doesn’t mean AI-generated writing in academia is dead. But it could mean that disclosing AI use in one’s writing is on the way to becoming the norm. And that, while not the ideal for many, is undeniably a win for Humans 1.0. Of course, for Humans 2.0 (or augmented humans, or transhumans) AI-generated text, disclosed or not, could become trivial in short order. Alas, for now, having a disclosure policy at least helps us know what we’re feeding our minds, and how much the chef (author) cares about what they’re serving us. ✍︎
New essays on AI, film, & entertainment innovation arrive by email. Subscribe to stay in the loop. All editorial text is written by humans.
Cover image: A modified screengrab from Dartmouth’s website page for Prof. Schnell via Dartmouth
