Student Uptake of GenAI

Many recent technologies have significantly shifted how humans interact with text: consider the shift from typewriting to word processing, or from visiting a library in person to searching online for sources. In many ways, GenAI is just the latest in a long series of evolutions of writing technologies (Narayanan & Kapoor, 2025). At the same time, it’s hard to underestimate Generative AI’s impact on higher education. According to the AI in Higher Education Global Survey 2026, most higher education faculty worldwide believe AI will bring “significant” or “transformative” change to the role of an instructor (p. 43), and the vast majority of students worry that their programs are outdated (p. 25) and are skeptical that faculty are prepared to guide their use of AI (p. 28). More fundamentally, GenAI tools have shaken one of our persistent beliefs about writing: that writing implies writers, human actors engaged in expression.

While the implications of this change are far-reaching, in this resource, we focus on the specific implications for undergraduate classrooms, exploring how undergraduate students think about and use AI tools.

What prior experience do students have with AI tools?

Many traditional-age undergraduate students have been using AI since high school or earlier, both in their personal lives and in educational settings. Pew Research (2026c) found that most (64%) US teenagers use AI chatbots, with just under 30% using them daily. The most common uses are searching for information and getting help with schoolwork, and over 40% have used AI tools to summarize materials such as articles or books (Pew Research, 2026c). Use among young people is increasing rapidly, with the Rand Corporation (2026) finding that most middle schoolers now report using AI to help with homework.

At the same time, prior GenAI exposure in educational settings varies widely: research from the Rand Corporation (2025) found that adoption in K-12 was uneven across subjects, grade levels, and unequally resourced school districts. Some schools are working to restrict the use of GenAI tools and technologies, but others are fully integrating AI tools such as Khanmigo in their curricula and regularly incorporating chatbots for personalized classroom activities. Faculty at the college level are similarly divided (Digital Education Council, 2026, p. 22).

Young people’s comfort with emerging GenAI technologies compounds existing generational differences in technology use and information-retrieval practices. Many young people have written almost exclusively on Chromebooks or computers, not paper, throughout their schooling; more than half of U.S. high school students spend the majority of class time on school-provided devices, with 29% using them 75% of class time or more (Fordham Institute, 2026). Members of Gen Z are also more likely than previous generations to encounter information pushed to them through algorithms (a “push” information ecosystem) than to find it through searching (Hassoun et al., 2023). These differences in writing and research processes shape students’ responses to texts and assignments in college classrooms, as well as to emerging GenAI tools.

How do college students use GenAI tools?

Tracking students’ use of GenAI is notoriously challenging because usage rates change rapidly and researchers must rely on students’ self-reports, but it’s clear usage rates are high and rising. For example, in one December 2025 survey of over 20,000 students worldwide, 65% of undergraduate student participants in the US and Canada reported using GenAI tools in their learning—notably lower than other regions surveyed, which ranged from 78% to 92% adoption rates (Digital Education Council, 2026, p. 21). Many faculty believe that true rates of GenAI use among students are even higher (e.g., Fitzpatrick, 2026).

Academic tasks are the most common use of GenAI among students (67%) (Chegg, 2025). Top uses of GenAI tools include searching for information (58%), brainstorming ideas (54%), refining written work (49%), learning new concepts (45%), drafting work (43%), and summarizing documents (43%) (Digital Education Council, 2026, p. 38).

The availability of GenAI tools has also shifted the ways students get support for their learning: GenAI is now the most common ‘first stop’ for struggling students, with 29% reporting they turn to GenAI tools before any other resource when “stuck on a concept or assignment” (up from just 10% in 2023). In contrast, only 24% of their peers first checked free online resources, 15% consulted friends or peers, 14% reviewed course materials, and just 8% reached out to professors or teaching assistants (Chegg, 2025).

Commonly cited reasons for relying on GenAI tools include convenience and a sense of psychological safety (Marsh et al., 2024). Students are also especially likely to use GenAI tools in courses they perceive as unrelated to their interests and on assignments they see as busywork (Heinen & Young, 2025; Marsh et al., 2024).

For those of us at UMN, data from the 2024 Student Experience at Research Universities (SERU) survey provide some insights into UMN students’ GenAI use by population. For instance, responses to questions about GenAI use vary significantly by major—for instance, only 10% of Carlson students reported that they’d never used GenAI tools, compared to 50% of nursing students. 25% of international students, many of whom speak languages other than English, said they used GenAI for language translation, compared to only 7% of domestic students. While the exact numbers may not be accurate due to underreporting and the passage of time, understanding trends is still useful, so it’s worth spending a few minutes exploring the disaggregated responses for the student populations you teach.

AI use is also not monolithic. Although some faculty may imagine students simply copy-pasting answers from a GenAI chatbot, this use is probably not the most common. As reflected in UMN SERU data, GenAI tools play a variety of collaborative roles in students’ writing: suggesting and summarizing sources, assisting in brainstorming and outlining, explaining difficult course material, providing general feedback or line edits, generating images or sound, writing and debugging code, and more. Sophisticated users of GenAI tend to prompt more carefully, use a variety of GenAI tools throughout the writing process, and blend human- and AI-generated text. Experienced users are also more likely to employ paid models that produce higher-quality outputs.

How do students feel about GenAI tools?

UMN students are confident in their knowledge of GenAI tools, agreeing in the 2024 SERU that they understand how AI generates responses (87%), understand how to effectively prompt (87%), and understand how the use of GenAI might enhance (95%) or detract from (94%) their learning. They also report understanding when it is appropriate to use GenAI for coursework (97%) and that instructors have discussed when it is appropriate to use GenAI to complete their schoolwork (92%).

At the same time, at least some UMN students—even those who use GenAI—have concerns about GenAI's impact on their learning and society at large (Haney, 2026). Their uncertainty reflects broader trends in opinions about GenAI, especially among young people. As of August 2026, 52% of Americans, and 55% of those under 30, reported they were “more concerned than excited about the increased use of AI in daily life”—a marked increase from similar polling just a few years ago (Pew Research, 2026d). 48% of American adults under 30 think GenAI will negatively impact society (Pew Research, 2026b), and job loss is a particular concern, with 41% of college students concerned GenAI will reduce job opportunities in their fields by the time they graduate (Digital Education Council, 2026, p. 31). In the college context, 66% of college students and 73% of faculty worry that “AI could make learning too shallow and discourage critical thinking and creativity” (Digital Education Council, 2026, p. 12), and 95% of faculty worry students will overrely on GenAI tools (AAC&U, 2026). Students also report concerns about hallucinations, data privacy, and other drawbacks of GenAI systems (Chegg, 2025). These concerns can provide starting points for critical conversations about GenAI tools in your courses.

At the same time, the sheer proliferation of GenAI tools can make them feel hard to avoid. Even students who worry about the impact of GenAI tools on learning and say they would support universities restricting GenAI use still often use these tools at least occasionally (Digital Education Council, 2026, p. 27). Faculty note that even students who want to avoid GenAI find it hard to resist (Sharlet, 2026), and many students easily rationalize GenAI use (Fitzpatrick, 2026). With GenAI tools being used widely and increasingly present in much of our software, those who want to avoid using GenAI must exercise constant, active willpower: no out-of-sight, out-of-mind here. To help students avoid overrelying on GenAI tools, scaffold critical engagement with GenAI tools and discourage unauthorized GenAI use.