The Evolving Role of Artificial Intelligence in Learning, Retention, Course Design, and Assessment

Artificial intelligence (AI) is fundamentally reshaping how learners engage with knowledge, think critically, and participate in academic work. Rather than functioning as a simple answer-generator, contemporary generative AI tools operate as dialogic learning partners that support inquiry, reflection, and co-construction of ideas. For example, in a secondary-school study, students used ChatGPT to ask questions, verify claims, and extend their interpretations of literature, which strengthened their epistemic engagement and promoted deeper critical thinking (Tang et al., 2024). When teachers positioned AI as part of an ongoing intellectual conversation rather than a shortcut, students learned to interrogate its responses and situate its insights within their own reasoning. This shift cultivates autonomy and encourages students to move beyond passive learning. However, the integration of AI into learning environments also brings important ethical and cognitive concerns. Large-scale models can produce contradictions, hallucinations, or incomplete representations of knowledge, requiring educators to teach students how to navigate AI critically and ethically (de Leon et al., 2025). Without careful design, AI can undermine learner trust, autonomy, or fairness. Thus, the impact of AI on learners is dual: it can amplify meaningful engagement, creativity, and higher-order thinking, while simultaneously requiring intentional structures to prevent misinformation and over-reliance (Shaidullina et al., 2024). 

Alongside transforming classroom learning, AI is also influencing student retention in online courses. Retention is deeply connected to how consistently and meaningfully students interact with digital learning environments. Studies using AI-supported learning analytics reveal that students who log in frequently, access weekly materials, interact with resources, and complete tasks on schedule are significantly more likely to stay enrolled and succeed (Akçapınar et al., 2019). These behavioral markers—such as total reading time, number of sessions, and steady engagement with online content—provide early indicators of both motivation and risk. Similarly, increased engagement with digital resources, including repeated logins and ongoing assignment submissions, has been identified as a strong predictor of persistence (Jokhan et al., 2022). When learners disengage, skip early tasks, or show inconsistent participation, their risk of withdrawal increases sharply. AI-powered early-warning systems help instructors identify these patterns as early as Week 3 or Week 6, enabling timely feedback, outreach, and targeted support. In this way, AI becomes a tool not only for detecting emerging barriers but also for strengthening the ongoing communication and structure necessary for student success. 

Effective online learning design plays a critical role in both engagement and retention. The Universal Design for Learning (UDL) framework offers a powerful model for creating accessible and inclusive online environments, especially for students with diverse learning needs. By emphasizing multiple means of engagement, representation, and action and expression, UDL provides instructors with a blueprint for designing instruction that reduces barriers before they arise (Flanagan & Morgan, 2021). Clear navigation, consistent organization, multimodal resources, and scaffolded directions are particularly important in digital spaces where students must self-regulate their learning. These design choices not only increase accessibility but also help learners manage cognitive load, sustain interest, and build confidence—factors known to support retention in online contexts. Complementing UDL, the Instructional Design for Academic Thriving (IDAT) model extends the conversation by linking course design directly to student wellbeing. Situated within the Community of Inquiry (CoI) framework, IDAT highlights how teaching presence, cognitive presence, and social presence can foster academic determination, meaningful processing, and self-regulated learning (Jaleel et al., 2024). Structured pathways, personalized feedback, opportunities for collaboration, and analytics-supported reflection all contribute to environments where students not only persist but thrive academically. 

As AI becomes a more visible presence in education, assessment practices must also evolve. Pearce and Chiavaroli (2023) argue that generative AI challenges the value of traditional written assessments because AI systems can now perform convincingly on many academic tasks, including high-stakes examinations. They propose a deliberate distinction between “assisted” assessments, where the use of AI mirrors real-world professional practice, and “unassisted” assessments designed to measure independent reasoning, judgment, and higher-order thinking. This approach aligns with UDL principles, which emphasize offering multiple ways for students to demonstrate their learning and ensuring that assessments remain accessible and meaningful (Flanagan & Morgan, 2021). When thoughtfully integrated, AI can support learning by helping students generate ideas, check their understanding, and receive formative feedback without replacing essential cognitive work. The IDAT model reinforces this by showing how assessments, like other course components, must support self-regulation, feedback processing, and deep engagement (Jaleel et al., 2024). Incorporating AI into assessments, therefore, is not simply about allowing new tools but about designing systems that strengthen both human capacities and digital literacy. By blending AI-assisted tasks with opportunities for unassisted demonstration of expertise, educators can build assessment ecosystems that maintain integrity, encourage critical thinking, and reflect the realities of modern learning environments. 

References  

Pearce, J., & Chiavaroli, N. (2023). Rethinking assessment in response to generative artificial intelligence. Medical Education, 57(10), 889–891. https://doi.org/10.1111/medu.15092 

Flanagan, S., & Morgan, J. J. (2021). Ensuring access to online learning for all students through Universal Design for Learning. Teaching Exceptional Children, 53(6), 459–462. https://doi.org/10.1177/00400599211010174 

Jaleel, B., Horsley, S., & Atkins, M.-A. (2024). Academic thriving and online course design: A conceptual model. International Journal for Academic Development, 29(3), 420–433. https://doi.org/10.1080/1360144X.2023.2221225 

de Leon, J., de Leon-Martinez, S., Artés-Rodríguez, A., Baca-García, E., & De las Cuevas, C. (2025). Reflections on the potential and risks of AI for scientific article writing. Actas Españolas de Psiquiatría, 53(2), 433–442. 

Tang, K.-S., Cooper, G., Rappa, N., Cooper, M., Sims, C., & Nonis, K. (2024). A dialogic approach to transform teaching, learning & assessment with generative AI in secondary education. Pedagogies: An International Journal, 19(3), 493–503. 

Shaidullina, A., Han, J., Zhunussova, A., Nugumanova, S., & Jumabekov, A. (2024). Ethical artificial intelligence for teaching-learning in higher education. Education and Information Technologies.