The rapid expansion of generative artificial intelligence (AI) in higher education is reshaping how students learn, create, and engage with academic tasks, prompting new opportunities and challenges for higher education (Dwivedi et al., 2023; Meyer et al., 2023). Tools such as ChatGPT are increasingly embedded in students’ study routines, supporting drafting, revision, ideation, and concept development (Balabdaoui et al., 2023; Peters et al., 2023). This study examines an instructional design in which undergraduate business students used ChatGPT to obtain formative feedback on team-based blog assignments in an Introduction to Organisational Behavior course. By analysing 19 complete team submissions—including drafts, ChatGPT feedback, revised blogs, reflections, and instructor comments—the study explores how students engaged with AI feedback, the types of revisions they implemented, and the learning processes evident in their interactions.

The results indicate that ChatGPT consistently produced structured feedback aligned with assignment criteria, offering suggestions related to clarity, argument flow, scenario development, visual presentation, and surface-level writing quality (Kasneci et al., 2023a; Zhai, 2022). When students used the instructor-provided prompt, the feedback more often addressed theoretical accuracy, relevance of examples, argument strength, and conceptual alignment with organisational behaviour frameworks. This aligns with research showing that prompt design strongly shapes the quality and specificity of AI-generated feedback (Eager & Brunton, 2023; Wang & Li, 2024). In contrast, self-generated prompts produced more superficial responses focused on grammar, tone, and readability.

Despite receiving conceptual feedback, most teams primarily made surface-level revisions such as adjusting formatting, reorganising paragraphs, refining headings, and adding images—changes that improve readability but do not deepen theoretical insight. Students seldom incorporated suggestions requiring conceptual effort, such as elaborating on organisational behaviour theories, strengthening scenario-theory alignment, or clarifying causal logic. This pattern reflects broader concerns that students often use AI for low-effort tasks rather than deeper reasoning (Balabdaoui et al., 2023; Kasneci et al., 2023b).

A notable finding is students’ limited use of iterative prompting. Although ChatGPT’s affordances are maximised through iterative engagement (Kasneci et al., 2023a; Peters et al., 2023), teams rarely asked follow-up questions. When they did, these queries revolved around grammar checks rather than probing theoretical understanding or reasoning processes. This suggests gaps in both AI literacy and feedback literacy—skills essential for interpreting and applying feedback effectively (Carless & Boud, 2018; Pinto et al., 2023).

Student reflections further revealed limited metacognitive depth. Most reflections described superficial changes, often in checklist form, with minimal explanation of the learning value of revisions. Few reflected on theory comprehension, argument structure, or conceptual refinement. This echoes research showing that when students lack feedback literacy, they struggle to engage critically with comments—whether from instructors or AI tools (Cotton et al., 2023; Obenza et al., 2023). Instructor comments reinforced these gaps: instructors frequently highlighted conceptual weaknesses that ChatGPT had also identified but students had not addressed.

Taken together, the findings reveal both the promise and limitations of AI-mediated learning. AI supports improved clarity, structure, and visual design—key aspects of communication in management education (Peters et al., 2023). However, deeper conceptual learning depends heavily on student prompting skills, evaluative judgment, and willingness to engage iteratively with feedback, confirming recent concerns about overreliance on AI and shallow engagement (Chaudhry et al., 2023; Dergaa et al., 2023). More broadly, the study underscores the emerging need for structured AI literacy and prompt-design instruction within business curricula (Dwivedi et al., 2023; Pinto et al., 2023).

This qualitative study was conducted in an online Introduction to Organisational Behavior course enrolling 65 undergraduate business students at a large open-enrolment U.S. university. Given the diverse academic backgrounds and digital experiences typical of such settings, the course provided an ideal environment for examining AI-supported formative feedback (Balabdaoui et al., 2023; Obenza et al., 2023). Students were assigned to 12 teams and completed multiple blog-style management challenge assignments requiring application of organisational behaviour theories to workplace scenarios. Two of these assignments incorporated structured AI-mediated feedback. Students created draft blog posts and were required to submit them to ChatGPT using an instructor-provided prompt aligned with assignment criteria, including clarity, theoretical accuracy, structure, and design considerations. Consistent with research on prompt engineering in education (Eager & Brunton, 2023; Wang & Li, 2024), this prompt was designed to elicit feedback relevant to writing quality and conceptual understanding. Students were encouraged—but not required—to ask follow-up questions to refine or deepen the feedback. After receiving ChatGPT’s feedback, teams revised their work and submitted a final package including: (1) the revised blog URL, (2) the original draft prompt and ChatGPT’s full response, and (3) a reflection explaining the revisions made and how the AI feedback was used. Instructor comments on the final submissions were also included. Three submissions were excluded due to inaccessible Google Sites pages, resulting in 19 complete datasets for analysis. The analysis proceeded in three stages consistent with thematic qualitative approaches (Kasneci et al., 2023a; Peters et al., 2023). First, ChatGPT-generated feedback was coded into categories aligned with assignment criteria: theoretical application, clarity and flow, design elements, call to action, and grammar/mechanics. This allowed comparison across teams using the instructor prompt versus self-generated prompts. Second, revised blogs were compared to the AI feedback to identify which suggestions students implemented and the depth of those revisions. Following research distinguishing surface-level from conceptual revisions (Carless & Boud, 2018; Kasneci et al., 2023b), changes were categorised as structural, stylistic, or conceptual. Third, student reflections were analysed for specificity, rationale, metacognitive insight, and evidence of evaluative judgment. Instructor comments served as a triangulating data source, revealing cases where conceptual feedback—provided by ChatGPT—had been overlooked. This multi-layered approach enabled a nuanced understanding of how students engaged with AI-mediated feedback and what these interactions revealed about AI literacy, feedback literacy, and learning processes in an authentic educational setting.

This study demonstrates that while generative AI offers meaningful support for improving clarity, organisation, and visual presentation in student writing, its capacity to foster deeper conceptual learning is limited without intentional instructional scaffolding. Students overwhelmingly adopted surface-level suggestions such as formatting and wording changes, reflecting patterns observed in recent research showing that students gravitate toward low-effort revisions when using AI (Balabdaoui et al., 2023; Kasneci et al., 2023b). Despite receiving accurate conceptual feedback, many teams did not revise theoretical explanations or strengthen scenario-to-theory alignment, suggesting gaps in both AI literacy and disciplinary understanding. The quality of prompts played a decisive role in shaping the usefulness of AI feedback. Teams who used the instructor-provided prompt received richer, targeted feedback aligned with learning outcomes, reinforcing emerging evidence that effective prompting is a core academic skill in AI-supported learning environments (Eager & Brunton, 2023; Wang & Li, 2024). By contrast, self-generated prompts yielded generic responses that did little to support conceptual refinement—a finding that underscores the importance of explicit prompt-design instruction within management education. Students’ limited use of iterative prompting further constrained the depth of learning. Although iterative engagement is central to maximising AI’s instructional value (Peters et al., 2023; Kasneci et al., 2023a), follow-up questions were rare and superficial. Reflections revealed similarly limited metacognitive engagement, aligning with feedback literacy research showing that students often struggle to interpret and apply feedback of any kind (Carless & Boud, 2018; Cotton et al., 2023). Instructor comments confirmed the need for continued human mediation: conceptual gaps overlooked by students were frequently identified by instructors, indicating that AI feedback alone cannot ensure disciplinary accuracy (Mollick & Mollick, 2022; Masters, 2023). Overall, the findings point to the need for integrating AI literacy, reflective judgment, and structured feedback processes into management curricula to ensure responsible, equitable, and pedagogically meaningful use of AI.

Balabdaoui, F., Szoniec, K. S., Bassetti, F., Dhaene, S., & Gruber, P. (2023). A survey on students’ use of AI at a technical university. Discover Education, 3(51). Chaudhry, I. S., Aziz, S., Bokhari, S. A. A., Sarwar, A., & Hussain, M. (2023). Time to revisit existing student performance evaluation approach in higher education sector in a new era of ChatGPT. Cogent Education, 10(1), 2210461. Cotton, D. R., Cotton, P. A., & Shipway, J. R. (2023). Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. Innovations in Education and Teaching International, 61(2), 228–239. Dergaa, I., Abubaker, M., Mataruna-Dos-Santos, L. J., Musa, S., & Zmijewski, P. (2023). From human writing to artificial intelligence generated text: Examining the prospects and potential threats of ChatGPT in academic writing. Biology of Sport, 40(15), 615–622. Dwivedi, Y. K., Kshetri, N., Hughes, L., Slade, E. L., Kar, A. K., Baabdullah, A. M., Koohang, A., Raghavan, V., Albanna, B., & Singh, S. (2023). “So what if ChatGPT wrote it?” Multidisciplinary perspectives on opportunities, challenges and implications of generative conversational AI. International Journal of Information Management, 71, 102642. Eager, B., & Brunton, R. (2023). Prompting higher education toward AI-augmented teaching and learning practice. Journal of University Teaching and Learning Practice, 20(5). Kasneci, E., Sessler, K., Kitchen, A., Kasneci, G., & Rittberger, M. (2023). AI-based feedback for higher education: Opportunities and challenges. Educational Technology Research & Development, 71, 1–23. Masters, K. (2023). Ethical use of artificial intelligence in health professions education. Medical Teacher, 45(6), 574–584. Meyer, J. G., Zhu, L., Varma, R., & Shapiro, M. (2023). ChatGPT and large language models in academia: Opportunities and challenges. BioData Mining, 16(1), 20. Obenza, B. N., Hilario, S. P., Cruz, K. P., & Malolos, R. C. (2023). University students’ perception and use of ChatGPT. International Journal of Human Computing Systems, 5(12). Peters, M. A., Rizvi, F., Gorur, R., & Zipin, L. (2023). Learning with ChatGPT 3.5 as a more knowledgeable other. International Journal of Educational Technology in Higher Education, 21(10). Pinto, A. S., Moreira, F., Reis, A., Carvalho, C. V., & Figueiredo, M. (2023). How machine learning is transforming higher education. Journal of Information Systems Engineering and Management, 8(2), 21168. Wang, L., & Li, W. (2024). The impact of AI usage on university students’ willingness for autonomous learning. Behavioral Science, 14(10), 956.