Institutions should move toward deliberate, transparent policies governing AI use in admissions and reconsider what application essays should assess¶

Strategy · All strategies
Evidence · 1 claim (1 mixed) · 1 study (1 causal), q2 · 0 of 1 report an effect size · 1 claim rests on one study

Description¶

Based on evidence that AI use is widespread, quality-improving, and penalized, the authors recommend that institutions "move toward deliberate, transparent policies around AI use", noting that many institutions lack policies and that where restrictions exist, boundaries of permissible assistance may be unclear and enforcement inconsistent. They argue universities must clarify what they seek to learn from application essays and reconsider whether essays' historical purposes remain appropriate when AI-assisted writing is becoming standard professional practice.

Design Implications¶

Context¶

Requirements¶

• Institutional commitment to defining the role of AI-assisted writing and to making policies explicit and enforceable

Constraints¶

• The authors state their empirical results on their own do not determine what role AI-assisted writing should play in a well-designed admissions process

Target Learners¶

• graduate and undergraduate program applicants
• admissions officers and administrators

Target Learning Goals¶

• fair and valid evaluation of applicant writing in an AI-mediated environment

Claims¶

• Admissions officers can often discriminate AI-written from human-written essays, achieving an AUC of 0.70, well above chance but below commercial detectors [~W]

Related Strategies¶

• Institutional governance should establish practical conditions supporting AI literacy, learner agency, and critical evaluation rather than relying on prohibition and detection
• Strengthen institutional policies, ethics education, verification, disclosure, and data protection to support ethical GenAI use

Examples¶

-

Key Sources¶

• Calvin Isley, Johann D. Gaebler, and Sharad Goel. (2026). AI-written admissions essays are widespread but penalized. arXiv preprint. https://arxiv.org/abs/2609.22549