Is It Ethical for Teachers to Use AI for Student Feedback?

Synthesis: Luo and Eaton (2026) examine whether it is ethical for teachers to use AI to provide Feedback to students, an important yet time-consuming part of teachers' work now increasingly automatable through Generative AI. Addressing a gap in empirical research on the ethical implications of this practice, the authors draw on a diverse dataset — university AI policy reviews from 50 leading institutions, Reddit social media posts, and interviews with university teachers and students in Southern China — to extrapolate eight key areas of ethical consideration for teachers' use of AI for feedback. Rather than offering a simple binary answer, they call for a nuanced understanding that weighs contextual complexities such as assessment purposes, assignment types, and the kinds of feedback being automated, and emphasize designing AI-enabled feedback activities that preserve the care, trust, and human connections central to effective feedback processes in higher education.

Key Findings

• There is no simple binary answer to whether it is ethical for teachers to use AI for student feedback; ethical use depends on multiple contextual factors including how AI use is communicated, assessment purpose, assignment type, student privacy, and teacher-student trust.
• The study identifies eight key areas of ethical consideration: (1) professionalism (AI use may conflict with teachers' professional responsibilities and erode professional identity and critical-evaluation skills), (2) accountability (teachers must remain answerable for and exercise human oversight over AI-generated feedback), (3) transparency (teachers should clearly communicate how AI is used and obtain student consent before uploading work), (4) effectiveness (impacts on teacher-student relationships, feedback quality, and teachers' instructional insight), (5) inclusivity (systemic AI bias, uneven application across students, but also potential to soften feedback language), (6) security (use of secure AI platforms and protection of student data and intellectual property), (7) contextual (assessment purpose, assignment complexity, and feedback type shape ethicality), and (8) policy and resource (alignment with institutional governance, access to secure platforms, and training).
• Transparency and consent emerged as central: eight universities advised transparency about AI use, four required student consent, and informed consent itself carries workload costs, leading some to propose "passive consent" (opt-out) models.
• Teacher-student trust and relationships are a recurring concern, with students expressing that AI-generated feedback can feel "disingenuous," eroding the trust essential to meaningful feedback uptake.
• Feedback as a relational practice: the authors frame feedback not as one-way information transfer but as a dialogic, relational practice, arguing effective feedback relies on human connections that AI may struggle to replicate.
• Policy gaps: only 14 of 50 top universities had specific guidance on teachers' AI use for feedback, and interviewed teachers reported little awareness of relevant internal policies — signaling a need for clearer governance and teacher training.

Study Design & Method

The study addresses the research question "What are the key ethical considerations for teachers when using AI to provide students with feedback?" using four datasets: (1) institutional AI policies and guidelines from 50 leading universities (2025 QS rankings), searched Jan–Feb 2025, with only 14 found to have specific feedback-related guidance; (2) Reddit social media posts — three topic posts on teachers' AI use for feedback (including one with 791 responses), yielding 76 analyzed responses; (3) interviews with 33 teachers; and (4) interviews with 23 students from Southern China institutions, where "AI+Education" national policy encourages AI integration. Purposive sampling ensured diverse disciplines. Drawing on Mahony's (2009) definition of Ethics as "what we ought to do," the analysis used open coding in NVivo, aggregating meaning clusters into eight overarching themes that were triangulated across datasets (all themes supported by at least two data sources; half by all four). Ethical approval was obtained (Approval Nos. 2023-2024-0134 and 2024-2025-0174).

What this means for practice

• Instructors. Communicate how AI is used in producing feedback and obtain student consent before uploading their work: eight of the 50 reviewed universities advised transparency and four required consent, and students described unannounced AI feedback as "disingenuous."
• Instructors. Retain accountability and human oversight — review AI-drafted comments against the assignment's purpose and the individual student before returning them, since answerability for feedback remains the teacher's.
• Instructors. Apply the same transparency standard to your own AI use that you expect of students; the study finds teachers who scrutinize student AI use are often less reflective about the ethics of their own.
• Administrators. Fill the guidance gap: only 14 of 50 top-ranked universities had specific policy on teachers' AI use for feedback, and interviewed teachers were largely unaware of the policies that did exist. Fund secure, institution-approved platforms, professional development, and a workable consent model (opt-out "passive consent" was proposed precisely because requiring informed consent adds workload).
• Designers. Build transparency and consent workflows into feedback tools and preserve space for teacher judgment across the study's eight areas (professionalism, accountability, transparency, effectiveness, inclusivity, security, contextual fit, and policy and resources). Note the paper's CC BY-ND (no derivatives) license when adapting its materials.

Limitations

• The eight areas reflect a purposive sample — AI policies from 50 elite (2025 QS-ranked) universities, 76 analyzed Reddit responses, and interviews with 33 teachers and 23 students in Southern China institutions — and the authors note that voluntary social-media participation, teachers recruited from the researchers' professional networks by snowballing, and voluntary interviews likely drew respondents with strong views on AI. The authors also flag that the policy sample may be biased toward Western elite perspectives.
• No demographic data were collected on participants' AI literacy, existing perceptions of feedback, or beliefs about AI use, so the authors cannot attribute differences to participant characteristics; the policy data itself covers only elite institutions.
• Interviews capture what teachers say, not what they do: the authors acknowledge that interview data may not reflect the actual decision-making teachers navigate when using AI for feedback.
• The study omits other stakeholders such as developers of specialized AI-powered feedback systems, so how ethical considerations feature in tool design is inferred rather than observed, and the policy landscape is changing fast enough that the 2025 snapshot dates quickly.

Citation

Luo, J., & Eaton, S. E. (2026). Is It Ethical for Teachers to Use AI for Student Feedback?. Journal of University Teaching and Learning Practice. https://doi.org/10.53761/887m5346. CC BY-ND 4.0.