Dartmouth provost in a real bind over use of artificial intelligence

The Globe’s reporting on Dartmouth’s provost and the use of artificial intelligence in his academic writing highlights a growing misunderstanding about what AI-detection tools actually measure (“Provost’s use of AI angers campus: Dartmouth leader relied heavily on tool, review says,” Page A1, Sept. 26). These tools do not determine who wrote a text or whether its ideas were generated by a machine. Rather, they compare the writing to statistical patterns found in modern academic prose, which is now routinely edited with AI‑assisted tools such as Grammarly or language models used for grammar and flow refinement.

As scientific writing has become more standardized and syntactically polished, it increasingly resembles the stylistic profile that AI systems produce. Ironically, the clearer and more coherent the prose, the more likely detectors are to flag it as “AI-generated.” A scholar who writes his own papers and uses AI only to check syntax or logical flow may therefore be misidentified as having outsourced authorship, even when the intellectual work is entirely his or her own.

Academic integrity is essential. But resemblance in style is not evidence of misconduct. Before reputations are damaged, the public deserves a more accurate understanding of what AI detectors can and cannot tell us.

The details about Dartmouth College’s provost, Santiago Schnell, and his use of artificial intelligence in his writing are appalling. I am sick of people who have done wrong saying things like, “I am sorry that this was not my standard practice,” as Schnell did, saying, “I now realize I should always explicitly disclose use of AI in my public writing.”

The response from Sian Leah Beilock, president of Dartmouth, was mealy-mouthed: “The questions this moment raise go well beyond any one person. Thoughtful people on our campus can and should disagree about those norms, which are rapidly shifting.”

What the college’s provost has done is wrong. Someone needs to say so. Since AI won’t say so, and since the president of Dartmouth won’t say so, I will.