Feedback has long been recognised as a cornerstone of effective learning, enabling students to identify strengths and areas for improvement in their academic work. AI tools such as ChatGPT, Claude, and Copilot have altered the feedback landscape in higher education by offering unprecedented access to immediate feedback. However, their widespread adoption raises important questions about how students and educators can harness their potential effectively.

Understanding feedback literacy in the AI era

Feedback literacy encompasses students’ ability to seek, interpret, and act upon feedback constructively. This skillset has become increasingly critical as AI-generated feedback becomes more prevalent across university settings. Students demonstrate varying levels of feedback literacy, which significantly impacts their ability to engage productively with AI tools. Those with high feedback literacy approach AI systems strategically, critically evaluating responses, identifying potential limitations, and seeking clarification when needed. Conversely, students with limited feedback literacy may accept AI suggestions uncritically or fail to recognise when additional guidance is necessary. This variation in student capabilities has important implications for how educators structure feedback processes and support student development in their courses.

Benefits and Challenges of AI-Generated Feedback

Advantages of AI Feedback: AI feedback offers several notable benefits that address common barriers to student feedback-seeking behavior. The availability of immediate, non-judgmental responses can reduce anxiety associated with seeking feedback and encourage more frequent engagement with the revision process. Students often appreciate the consistent availability of these tools, which allows them to work at their own pace without scheduling constraints.

Significant Concerns: However, AI systems can produce biased, inaccurate, or misleading information, potentially misdirecting student learning. The tendency toward homogenised responses may limit exposure to diverse perspectives and approaches. Additionally, over-reliance on AI feedback can undermine students’ ability to think independently.

Research indicates that students continue to perceive human feedback as more valuable than AI-generated alternatives, highlighting the continued importance of educator involvement in the feedback process.

Complementary Feedback Systems

Given the reality that students are already incorporating AI tools into their learning, educators must focus on developing student competency in using these resources effectively while adhering to these three guiding principles:

1. Supporting students with developing their feedback literacy

Teaching students how to effectively seek, interpret, and act on feedback from both AI and human sources. This, for example, includes learning to craft better prompts for AI tools. The CLEAR framework provides a structured approach, emphasising prompts that are Concise, Logical, Explicit, Adaptive, and Reflective. Rather than generic requests, students should learn to ask specific questions that target aspects of their work. They should also understand when different types of feedback (peer, teacher, AI, self) are most valuable and how to use it constructively for improvement.

2. Building students’ evaluative judgement

Developing students’ ability to critically assess the quality, relevance, and accuracy of feedback they receive. This means helping them recognise when AI feedback might be biased or incomplete, when to seek clarification, and how to make informed decisions about which suggestions to implement in their work. Students need practice in evaluating feedback from different sources by deciding which recommendations to accept, reject, or modify, providing justification for their decisions.

3. Strengthening the human connection in the feedback process

Feedback is not only about information transfer but also about building human-to-human trust and respectful relationships. Human feedback remains irreplaceable for providing context, understanding individual student needs, and fostering the kind of dialogic engagement that promotes deep learning. The back-and-forth nature of human interaction allows for clarification and negotiation of meaning but also recognition of when students might need support, challenge or encouragement beyond what their work explicitly reveals.

If our hope is for students to use generative AI critically, with awareness and autonomy then as educators we need to design AI feedback practices in a way that intentionally helps students to develop these skills. To succeed in this task, we must engage with these systems while preserving the human elements that give feedback its educational power.

What next?

• Get inspired: read how a Cogniti agent has been used as a feedback assistant in essay plans
• Get hands on: register for the AI workshop lab on effective prompt writing
• Get deeper insight: register for the MPLF module on assessment and feedback for learning

About me

I am an academic developer within the Educational Innovation division of the DVC-Education Portfolio at the University of Sydney. With over 20 years of experience as a lecturer, academic English teacher, and curriculum designer, I have developed a strong interest in how both students and educators learn best. My PhD, completed at the Centre for Research in Assessment and Digital Learning (CRADLE) at Deakin University, explored how ipsative feedback processes provide learners with opportunities to foster distinctive feedback literacy competencies. These days, I am interested in how emerging AI tools might help us build better learning partnerships in feedback and assessment.