Dartmouth Provost: Universities Need AI Transcripts, Not Just Bans

Universities are deploying traditional anti-cheating measures to combat AI-generated student work, but these tactics address only surface-level problems according to a senior academic administrator.

Santiago Schnell, provost of Dartmouth College and a mathematics professor, argues that while institutions revive proctored exams, blue books, and oral assessments to verify authentic student learning, these methods fail to solve a fundamental challenge. The University of Chicago Law School exemplifies this approach with a pilot program this fall that will generally prohibit electronic devices in core first-year classes, with limited exceptions.

The transcript problem

These measures can successfully deter cheating and help faculty assess what students can accomplish without AI assistance. However, Schnell contends they don't address what he sees as the deeper issue: academic degrees currently fail to distinguish between capabilities students have developed independently and those they can only demonstrate with AI tools.

The provost's position, detailed in The Washington Post, suggests that simply preventing AI use during assessments leaves unresolved questions about how educational credentials should represent student competencies in an AI-augmented world.

Why it matters

Employers and graduate programs rely on degrees as signals of what graduates can do. If transcripts don't clarify which skills students mastered independently versus with AI assistance, credentials lose precision as hiring and admissions tools. This ambiguity could undermine the labor market value of higher education at a time when universities already face questions about return on investment.

Beyond policing

While Schnell acknowledges that preventing AI-assisted cheating remains important, his argument points toward a more complex reckoning for higher education. Rather than focusing exclusively on detection and prohibition, institutions may need to redesign how they document and certify learning outcomes.

The challenge extends beyond academic integrity to questions of credential design: Should transcripts indicate which courses permitted AI use? Should degrees certify both AI-assisted and independent competencies separately? These structural questions remain largely unaddressed as universities concentrate on immediate enforcement measures.

Schnell's perspective, first reported by The Washington Post, comes as institutions nationwide grapple with how to adapt assessment practices and academic policies to generative AI tools that became widely available less than two years ago.