Here is the uncomfortable truth about AI in admissions outreach: the technology is not the problem. The way many institutions are deploying it is.
Colleges racing to automate student communication are solving a capacity problem while quietly creating a trust problem—and trust is the only currency that matters in a decision as personal as where to spend four years of your life.
The 2026 Enrollment Engagement Report captures the tension precisely. Nearly half of students surveyed (47%) already use AI tools like ChatGPT or Gemini during their college search. Yet 83% say they would rather get answers from a real person, and nearly 60% report a more negative perception of a college when a message feels AI-generated.
Students are not anti-AI. They are anti-being fooled. They draw a sharp line between using AI as a research tool and receiving AI as a substitute for relationships, and they spot the substitution faster than most enrollment teams realize.
As the report puts it, “Students don’t want colleges to sound automated. They want colleges to notice them.”
This tracks with broader American attitudes toward AI. Edelman’s most recent research on AI and trust found that only about a third of Americans trust the technology, with nearly three times as many rejecting its growing use as embracing it.
And the 2026 Edelman Trust Barometer names generative AI among the top events eroding institutional trust, finding people retreating into smaller circles they already know and trust.
Colleges are not deploying automated outreach into a neutral environment. They are deploying it into a public primed for skepticism—one that increasingly reserves its trust for real human relationships—while asking families to make one of the most personal, most expensive decisions of their lives.
So the question for enrollment leaders is not whether to use AI. It is where.
Engagement hasn’t disappeared
The answer is to treat AI as an intelligence layer, not a voice. The enrollment research shows why this is urgent: 53% of students applied or requested information without any prior direct interaction with the institution.
Engagement hasn’t disappeared, it has gone quiet, moving into a “dark funnel” where students evaluate colleges long before a CRM ever sees them. This is where AI genuinely shines: interpreting behavioral signals, detecting emerging intent, scoring prospects on cumulative engagement rather than form submissions alone, and surfacing the students who need human attention right now.
Every hour AI absorbs in analysis and prioritization is an hour a counselor can spend on the phone. Used this way, AI doesn’t replace human connection, it promotes it.
Where AI should not live is in the moments students experience as relational, and students have told us exactly which moments those are.
Sixty-five percent prefer a real person for conversations about academic fit. Seventy-six percent prefer a person for questions about campus life. Nearly three in four want to talk with a counselor about scholarships and financial aid.
These aren’t information requests; they’re trust requests. A student asking about scholarships may really be asking whether she can afford to belong. AI can identify the topic. Only a trained person can hear the concern underneath it.
Let AI do what machines do well
Four practical steps follow. First, audit every student-facing message that is AI-generated or AI-assisted, and read each one as a seventeen-year-old would: does this sound like a person who cares or a system that processed a name?
Second, formally protect the high-trust moments (fit, campus life, financial aid, final decisions) from automation, and keep a named human attached to them. AI can draft and prepare; a counselor must own and deliver.
Third, move personalization earlier. Nearly two-thirds of students believe colleges should be able to infer their interests from engagement behavior before an inquiry form is submitted, and 83% are more likely to act on messaging that reflects what they’ve actually explored.
If your personalization begins at the form, you’re starting after many students have already made up their minds.
Fourth, measure trust, not just speed. Alongside response rates and deposits, track opt-outs, chatbot dissatisfaction, and requests to speak with a person. Those are early warnings that efficiency is costing you relationally.
The obvious objection is capacity. Admissions teams are stretched thin, and automating outreach feels like the only way to keep up. But that logic runs backwards.
Understaffed teams have the least trust to spare, and mass-produced “personalization” erodes it at scale. Smaller private institutions in particular recruit on the promise of being known—mentorship, community, personal attention. A generic AI-generated message breaks that promise before a counselor ever gets the chance to keep it.
The institutions that win the next decade of enrollment will not be the ones that automate the most messages, nor the ones that refuse the technology. They will be the ones that let AI do what machines do well—pattern recognition, prioritization, preparation—so their people can do what only people can do: make a seventeen-year-old feel that a real community, full of real humans, is genuinely glad they asked.
