Generative AI (GenAI) offers transformative potential for real-time feedback in education. By instantly analysing text, code, or responses, GenAI expands access to formative feedback, supporting active learning and self-reflection beyond class time or instructor availability. However, equitable implementation is essential: institutions must ensure access for all learners, provide training in digital literacy, and encourage critical engagement so that students evaluate rather than simply accept AI-generated feedback.

Expert oversight remains vital. While GenAI can identify structural or reasoning issues, it lacks the contextual understanding that human educators provide. It can also produce misleading or inaccurate outputs, so results must be critically reviewed. Secure, institutionally-managed systems are needed to safeguard confidentiality and data integrity, promoting ethical and sustainable use. Environmental sustainability should also guide adoption through efficient technologies and purposeful application.

In a course on optimisation, students compared analytical problem-solving using GenAI with independent numerical work. The exercise revealed insights into trust and feedback in learning. Some students trusted their skills and preferred independent work, while others discovered that effective GenAI use required careful prompting; vague inputs yielded generic results, but targeted ones encouraged critical thinking. Comparing GenAI’s analytical explanations with their own solutions became a key learning moment, reinforcing that GenAI outputs are hypotheses requiring verification rather than final answers. Though GenAI’s reasoning was often sound, occasional inaccuracies in its visualisations reminded students of the need for critical evaluation.

In another programming course, students were asked about using GenAI to assess progress on their projects while receiving timely feedback. Opinions were split: some valued GenAI’s structured, detailed approach that systematically addressed each point, while others appreciated the conversational and empathetic nature of peer feedback, which provided social connection, mutual support, and insight into classmates’ progress. Both forms of feedback were viewed as equally useful but for different reasons, as GenAI offered clarity, precision, and depth, while peers contributed context, collaboration, and a sense of shared learning.

The impact of AI feedback varied by subject. In mathematical contexts, precision and logical validation and accuracy were important; in programming project progress, clarity of explanation, creativity and collaboration amongst students mattered more.

Ultimately, when used thoughtfully, transparently, and under expert oversight, GenAI can democratise access to timely, personalised feedback for diverse learners. It serves as a sustainable and adaptive partner in learning: enhancing reflection, equity, and understanding while complementing, not replacing, meaningful human engagement.

About the authors

Dr Gule Saman is an Associate Professor at Heriot-Watt University. Her research explores inclusivity, work-based learning, and active student engagement. Dr Hebatallah Shoukry is an Assistant Professor at Heriot-Watt University. Her research focuses on threshold concepts, work-based learning and academic collaboration to enhance student engagement.

Corresponding author: Gule Saman, s.gule@hw.ac.uk