The study, titled “Strategic Integration or Skill Compensation? Understanding GenAI Use in Online Higher Education,” was published on June 30, 2026, in the journal Artificial Intelligence in Education, appearing in volume 2, issue 1, on pages 205 through 223. It was conducted by Jessica Sylvester, Melinda Kulick and Leonidas Maganares, all holding doctorates of education, along with Stella Smith, PhD, associate university research chair. All four are fellows or scholars affiliated with the university’s Center for Educational and Instructional Technology Research, known as CEITR, which is housed within the University of Phoenix College of Doctoral Studies and supports research on teaching, learning, educational technology and institutional practice.

The research team surveyed 491 non-traditional undergraduate students enrolled in online general education courses at the university. Non-traditional students, broadly defined, are typically older than the conventional college-age cohort and often balance coursework with jobs, family obligations and other responsibilities, making them a population of particular interest as universities grapple with how artificial intelligence is reshaping learning. Of the 491 students surveyed, 43.2 percent, or 212 students, reported using generative AI tools in their coursework, a figure that indicates the technology has moved well past the novelty stage and into routine academic life for nearly half of the online learners surveyed.

To understand what drove that use, the researchers turned to multiple regression analysis, a statistical technique that allows investigators to assess the independent contribution of several predictor variables to a single outcome while holding the others constant. The analysis identified exactly two statistically significant predictors of how frequently students used generative AI. The first was grade point average: students with stronger academic performance reported using generative AI tools more often. The second was technology confidence: students who expressed greater confidence in their own technical abilities also reported more frequent engagement with the tools. Both associations point in a direction that challenges the compensation narrative that has dominated much of the public conversation about AI in higher education.

Just as revealing was what did not predict use. Confidence in reading, writing and mathematics, the core academic skills that many observers assumed would push struggling students toward AI assistance, showed no statistically significant relationship with how often students turned to generative tools. Nor did time pressure: the number of hours students spent studying, working or meeting family responsibilities was not significantly related to generative AI use. In other words, neither a deficit in foundational academic skills nor a crunch on available time, the two most commonly cited explanations for why students might lean on AI, emerged as a meaningful driver of behavior in this sample.

The performance data among users add further texture to the picture. Among the 212 students who reported using generative AI in their coursework, 91.1 percent held a GPA of 3.0 or higher, and 68.9 percent earned a perfect 4.0 grade in the specific course in which they reported using the technology. Whatever generative AI is doing in the lives of these learners, the study’s authors argue, the tools appear to be concentrated among students who are already performing well and who feel comfortable navigating digital environments, rather than among those who are falling behind.

“These findings complicate the assumption that students primarily turn to generative AI because they are struggling academically or simply looking to save time,” said Jessica Sylvester, senior manager of College Operations at University of Phoenix and lead author of the study. “Instead, we found GenAI use associated with stronger academic performance and digital confidence. That points to an important opportunity for higher education to help all learners develop the digital fluency, critical judgment and ethical awareness needed to engage with AI effectively.” Her remarks frame the results less as a celebration of AI adoption than as a signal that institutions may need to actively teach the skills that make AI use productive rather than substitutive.

Open-ended survey responses from the 212 students who reported using generative AI offered a more granular view of how the technology is actually woven into coursework. Three themes dominated. The most prominent was using AI to improve understanding and clarify complex concepts, suggesting that many students treat the tools as a personalized explainer rather than a text generator. The second theme cast generative AI as a “thinking partner,” a learning aid for brainstorming and organizing ideas before or during the writing process. The third was efficiency: students reported that the tools helped them work faster and be more productive on academic tasks. Taken together, the qualitative responses sketch a portrait of AI as an amplifier of existing study strategies rather than a replacement for them, at least among these high-performing online learners.

The students themselves were not uncritical enthusiasts. Alongside their enthusiasm, respondents raised concerns about the accuracy and originality of AI-generated content, worried about preserving their individual academic voice, and expressed uncertainty about what constitutes appropriate use under their institutions’ academic integrity policies. That last concern is a recurring tension in higher education, where policies on generative AI have often lagged behind adoption, leaving students to navigate ambiguous rules on their own. The researchers argue that this uncertainty itself is a problem worth solving, and that the findings support a shift away from approaches centered primarily on restricting AI and toward clearer guidance that cultivates generative AI literacy, critical evaluation skills, and transparent and ethical use.

The study carries specific practical recommendations for institutions. The authors suggest incorporating AI literacy across academic disciplines rather than confining it to technology courses, designing assignments that encourage students to critique and reflect on AI-generated information, and providing consistent institutional guidance about acceptable use. The rationale is straightforward: if the students most likely to engage with generative AI are those with the strongest grades and the greatest technology confidence, then leaving AI skills to develop informally risks widening a digital fluency gap between confident and hesitant learners, precisely the outcome an equitable higher education system would want to avoid.

The authors are careful to flag the limits of what their data can show. The research used a cross-sectional, correlational design and drew a non-probability sample from a single online university, which means the findings cannot establish that generative AI use causes stronger academic performance, and they should not be generalized to all college students. It is equally plausible that higher-performing, more tech-confident students are simply more willing to adopt the tools, or that the direction of the relationship runs the other way, or that unmeasured factors influence both use and performance. To untangle those possibilities, the authors recommend longitudinal and multi-institutional studies that track how patterns of generative AI use evolve over time and whether AI-supported learning affects longer-term academic outcomes.

Even with those caveats, the study arrives at a consequential moment. Universities worldwide are revising integrity policies, redesigning assessments and debating the role of AI in classrooms as adoption accelerates among students of every description. Evidence that frequent use among adult online learners is associated with stronger rather than weaker performance cuts against both alarmist bans and complacent embrace, and instead supports the more demanding path the authors propose: teaching students to use these tools thoughtfully, critically and ethically. Sylvester and Kulick previously presented findings from this research at the 2025 Association for Educational Communications and Technology International Convention, and a summary of that presentation, “Transforming Higher Education: Harnessing Generative AI for Student Engagement,” was subsequently published in The Journal of Applied Instructional Design, reflecting the team’s sustained effort to move the conversation about generative AI from panic and prohibition toward evidence-informed pedagogy. As the technology continues to evolve, studies of this kind, grounded in the actual behaviors of the students most affected, will be essential to ensuring that institutional policy keeps pace with classroom reality.

Subject of Research: Generative AI use and its association with academic performance and technology confidence among non-traditional undergraduate students in online higher education

Article Title: University of Phoenix study finds GenAI use among online adult learners associated with academic performance and technology confidence

Article References: University of Phoenix study finds GenAI use among online adult learners associated with academic performance and technology confidence. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: generative AI, online education, adult learners, academic performance, technology confidence, AI literacy, academic integrity, higher education, non-traditional students, University of Phoenix, ChatGPT, educational technology