50,000 Students and Faculty Reveal Higher Ed's AI Adoption Gap

Two of the largest higher education AI surveys ever conducted have landed within weeks of each other, and together they paint a troubling picture: students are adopting AI rapidly while institutions struggle to provide guidance, and faculty are retreating from AI instruction rather than leaning into it.

The Lumina Foundation–Gallup 2026 State of Higher Education study surveyed nearly 4,000 U.S. students, while the Digital Education Council's global survey captured 45,398 responses across 35 countries. The combined findings reveal what the Digital Education Council calls "adoption without authority"—students deeply embedded in AI use but operating without institutional frameworks, faculty preparation, or assessment alignment.

Why it matters

This gap has direct workforce implications. Nearly half of U.S. students have seriously considered changing their major because of AI's impact on the job market, and 16 percent have already done so. When institutions fail to provide AI guidance, students make their own adjustments—but without the structured learning that builds genuine competency. The result is a generation of graduates who use AI tools daily but lack the domain expertise and critical judgment employers will demand.

Three paradoxes defining the crisis

The survey data reveals three contradictions that explain why progress has stalled.

First, 43 percent of U.S. and Canadian students say they would not be disappointed by an institution-wide AI ban—the highest tolerance for restrictions of any region surveyed. In Latin America, only 15 percent feel the same. Yet 76 percent of students globally have never participated in any AI literacy training, and nearly half do not even know whether such training exists at their institution. Students at schools that ban AI are the most likely to feel undertrained—and the most likely to use it anyway. One in four students at institutions with outright bans still report weekly AI use.

Second, faculty are disengaging rather than adapting. Globally, 73 percent of faculty worry that students are using AI at the expense of developing their own skills. But in the U.S. and Canada, faculty intent to use AI in future teaching dropped nine percentage points in a single year—from 76 percent to 67 percent—the sharpest regional decline measured. This creates a self-reinforcing cycle: faculty pull back, students receive less guidance, and the skills erosion faculty fear becomes more likely.

Third, faculty and students on the same campuses are reaching opposite conclusions about curriculum relevance. Fifty-eight percent of U.S. and Canadian faculty say they are not worried that what they teach will be outdated by graduation—the highest faculty confidence of any region. Yet only 19 percent of U.S. and Canadian students believe their program feels current and relevant in terms of AI and future skills—the lowest student confidence globally.

The competency gap in practice

The surveys quantify what unguided adoption looks like: 72 percent of students globally say their assessments do not consistently reflect the work, skills, and judgment they will need in AI-enabled workplaces. In the U.S. and Canada, 48 percent say none or only a few assessments do. Only 29 percent of students globally believe their instructors are well equipped to guide them on AI use. In the U.S. and Canada, that figure drops to 17 percent.

The weakest dimension of student AI competency is domain expertise—knowing how to apply AI within a specific discipline. Only 35 percent of students can use AI tools for discipline-specific tasks and identify which tools work best and why. The strongest dimension is human-centricity: understanding when human oversight is needed. Students recognize that AI requires judgment; they simply have not been taught how to exercise it in context.

Institutions closing the gap

Some institutions are treating AI as a structural priority rather than a syllabus footnote. Purdue University's trustees approved an AI working competency as a graduation requirement. Georgia State University's GPS advising system analyzes 800 risk factors nightly for 50,000 students, improving six-year graduation rates by 23 percent. The University of Hawaii deployed AI chatbots across all 10 campuses, achieving 94 percent student opt-in and answering 4,294 questions without human intervention in five months. The Medical University of South Carolina grounded its AI literacy initiative in the Digital Education Council's AI Literacy Framework, building shared competency benchmarks rather than leaving faculty to improvise.

The findings were first reported by Dr. Aviva Legatt in Forbes, drawing on both the Lumina Foundation–Gallup and Digital Education Council survey data.