Patrick Wheeler is the Executive Director of the Center for Digital Strategies at the Tuck School of Business at Dartmouth College. The From Awareness to Agency: Practical AI for Enrollment Teams webinar series is presented in partnership with the Duolingo English Test team and runs across Summer 2026.
Most of the visible debate about AI in admissions is about the applicant side: How do we detect students’s use of AI? Is using AI to write an essay cheating? Is the essay dead? These are important questions, but they're downstream. The bigger question is what admissions offices are doing on their side of the table. Because while many admissions teams may already be using AI to do things like polish their emails or summarize lengthy documents, they’re not getting the full value out of this powerful tool.
Why most admissions teams are struggling to get value from AI
There’s a lot of hype about AI. But the data on the value most organizations are actually getting out of AI is sobering. And two of the most credible global studies on enterprise AI adoption are pointing at the same problem.
Boston Consulting Group's September 2025 Build for the Future study, The Widening AI Value Gap, surveyed more than 1,250 companies globally. It found that only 5% are achieving AI value at scale. Another 35% are scaling their efforts and seeing some returns, but most of those companies admit they are not moving far enough or fast enough. Sixty percent have little or nothing to show for their AI investment: minimal revenue gains, minimal cost gains, despite substantial spending. Of those laggards, over 30% admit they have made no progress at all.
These results are consistent with McKinsey's most recent State of AI report, which reveals that nearly two-thirds of organizations have not yet begun scaling AI across the enterprise. Most are still in experimentation or piloting.
Stated as one finding: the vast majority of organizations (95%, according to BCG) are not yet generating value from AI at scale. And because Admissions offices operate inside institutions subject to the same dynamics BCG documents across every sector studied, Higher Education is not exempt from these challenges. Recent reporting in Forbes and the Associated Press makes a few things clear: AI is no longer a horizon technology in college admissions. It is already inside the function. Some institutions are using AI tools to score essays. Some are running candidate-evaluation models that surface ranked recommendations to readers.
Now run that finding through admissions. A meaningful share of applicants are already using AI to draft and polish essays. If your office is using AI to score those same essays, you are not measuring applicants. You are measuring how well their AI tool stylistically matches yours. That is not a bias you can train your readers around or rubric your way out of. It is baked into the model. And it is exactly the failure mode you would expect when one technology is deployed on both sides of an evaluation it was never designed to mediate.
Where AI can create the most value in admissions
This research should land as a stop sign. But despite a growing body of evidence that the resulting decisions are partly an artifact of toolchain alignment rather than candidate merit, recruiters in the corporate world and admissions teams in higher education are continuing to deploy AI at the evaluation point, at scale.
The cleanest defense against this failure mode is not better calibration or a different model. It is not to deploy AI at the point of evaluation in the first place. Because the high-leverage uses of AI sit elsewhere.
AI is genuinely valuable in admissions not for evaluation itself, but for the complex analytical work that surrounds it: Class shaping under multi-constraint pressure; Yield modeling and personalized yield campaigns; Waitlist optimization against tight financial aid budgets; Interview design that probes the actual open questions in a file; Reader calibration on the coded language in recommendation letters; Equity-aware policy stress-testing before a decision gets institutionalized. These are the use cases I build into workshops I run for admissions teams. They are largely absent from current coverage, and they are the use cases that would actually improve outcomes.
If you work in admissions, it’s the difference between a team that is quietly using ChatGPT to clean up emails while the rest of the function gets harder, and a team that can run more rigorous yield analyses, evaluate authenticity questions with consistency, and pressure-test policy decisions before they hit the inbox.
The admissions offices that move from AI awareness to AI agency this summer will spend the next cycle solving better problems. The ones that don't will spend it explaining why they fell behind.
Moving from AI awareness to AI agency
I hear two important concerns from admissions professionals: That AI will erode the professional judgment that makes this work meaningful, or that it will be used as cover for thinning out the team.
Both fears are reasonable, and both describe the same misuse pattern I named above: AI deployed at the point of evaluation rather than in service of evaluation. That is the failure mode. Luckily, it is not the only path.
Agency, in the sense I mean it, is the capacity to direct an AI toward a decision your team actually owns. The decision still belongs to the human; the AI is a force multiplier for the analysis behind it.
This might look like an admissions committee modeling a waitlist pull across three constraints (regional gaps, STEM mix, first-gen percentage) and stress-testing it against the financial aid budget in 30 minutes instead of three days. Or a reader using AI to decode the coded language in a recommendation letter or surface inconsistencies across a file, without outsourcing the final judgment. Or a director pressure-testing an authenticity or AI policy by working through five hard edge cases with an AI partner before the policy goes to committee.
These are examples of more high-leverage ways your team could be using AI. Every session in this series is grounded in the principle that AI expands what your team can do, particularly the high-judgment, high-context work that is the whole point of the function.
A practical AI training series for admissions professionals
This summer, in partnership with the Duolingo English Test team, I'm running From Awareness to Agency: Practical AI for Enrollment Teams, a three-part webinar series built specifically for admissions and enrollment professionals who want to move from the wrong side of that value gap to the right side.
Session 1. Building Your AI Foundation
Before tools, context. The single biggest determinant of whether an AI output is useful is not which model you're using. It's the context you have given it. We'll work through how to build personal, team, and institutional context documents that make every subsequent AI interaction better. We'll also talk about how to bring along skeptical colleagues, what team norms for transparency look like, and how to land on the right side of the value gap structurally rather than through individual heroics.
Session 2. AI for Admissions Professionals
A hands-on deep dive into the questions you are actually wrestling with this cycle. How do you evaluate application authenticity in a world where most applicants have access to AI? What does an equity-aware AI policy for application review actually say? How do you use AI for cohort-level analysis and waitlist modeling without baking in last year's biases? How do you generate synthetic applicant data for training new readers? These are the questions I keep hearing in conversations with admissions leaders. Session 2 is built around them.
Session 3. Advanced AI: Agents, Automation, and Risk
For teams ready to move past prompting and into building. We'll cover what agentic workflows actually are, where they belong inside admissions operations and where they don't, how to manage cost and governance at the team and institutional level, and how to draft an AI adoption roadmap that survives leadership turnover. This is the session that takes you from "I use AI" to "we have an AI capability."
Join us. Bring a colleague. You can attend one, two, or all three. The exercises and materials are designed so any practitioner can join at any point. You will leave with materials, frameworks, and a peer network that changes how your team works the following week.