How should assessors use AI for marking and feedback?

Since the introduction of generative AI (GenAI) into higher education, the focus has been on its use by students in assessments. Many institutions have issued a blanket prohibition, even though GenAI can enhance feedback and marking for time-starved academics. 

At institutions where GenAI is permitted, assessors need to be transparent about their use of it and include students in the decision-making process. At my institution, we took this approach by running a half-day workshop with more than 40 participants, including students, assessors and professional services colleagues, to develop draft institutional principles for the transparent and pedagogically sound use of AI by assessors. This resource sets out draft principles we co-created.

Proposed draft principles:

1. Consistency in marking within the same assignment: the lead examiner is referred to as the “assessor” in these principles. The assessor is responsible for overseeing the overall module assessment and must ensure consistency in marking and feedback practices across all markers involved in the same module to promote transparency and alignment.

2. Purposeful enhancement of learning: should assessors decide on the use of AI, its application in assessment, marking and feedback must primarily enhance student learning experiences and improve the academic environment for all stakeholders. Its use should be grounded in evidence-based, ethical and responsible practices.

3. Unwavering human accountability: the ultimate responsibility for the quality, accuracy, fairness and pedagogical value of all assessment, marking and feedback remains with the human assessor. AI tools are aids that do not diminish or transfer this accountability. Challenges to academic judgement will not be accepted, in line with standard university regulations.

• Spotlight guide: AI and assessment in higher education
• AI and assessment redesign: a four-step process
• Designing assessments with generative AI in mind

4. Informed assessor autonomy and professional development: assessors retain autonomy in deciding whether and how to incorporate AI tools into their assessment, marking and feedback provision processes. They should exercise this autonomy with due consideration for pedagogical effectiveness, student equity and adherence to established best practices. It is the responsibility of the assessors and markers to familiarise themselves with best practices and the general capabilities of AI.

5. Transparency and trust: the integration of AI into assessment, marking and feedback provision must be fully transparent to build and maintain trust. Assessors must clearly disclose if and how they used AI tools, outlining specific methods and critically reflecting on strengths, limitations and potential biases. For students, this disclosure may be presented in summary form, while full details should be available for departmental review and external examiners.

6. Data privacy and security: strict adherence to data privacy, security and intellectual property protection is paramount when handling student work and personal data with AI tools. When uploading student work, assessors must use university-approved AI tools to ensure data protection.

7. Collaborative practice, equitable access and continuous adaptation: the university will actively foster a culture of open dialogue and collaborative exchange among staff regarding the effective, ethical and responsible use of AI in assessment, marking and feedback, while providing continuous professional development and support. All assessors and markers must have equitable access to necessary AI tools, as well as support and training in their use. The university commits to continuously evaluating and adapting its policies and practices regarding AI in assessment, marking and feedback practices, regularly reviewing the performance, efficacy and ethical implications of AI tools as technology evolves and new insights emerge.

While these are still only draft principles, the more institutions develop their own, the more acceptable it will become for assessors to use GenAI. If you are planning to develop your institutional principles in a similar multidisciplinary workshop set-up, here is what worked well for us: 

• Start with introductory presentations
• In smaller breakout groups, review and comment on the pre-drafted proposed principles (I recommend forming student-only groups with members from different departments, so that the subsequent plenary discussions can include clearly identifiable student voices. Avoid mixing students and staff in the breakout groups to preserve this distinction.)
• Breakout groups provide their feedback in a plenary discussion, keeping discussions focused on the underlying sentiment and values of the principles rather than precise wording
• Finalise the exact wording of the proposed draft principles post-workshop

I would like to thank colleagues from King’s College London, the London School of Economics and Political Science and the University of Southampton whose draft principles informed our half-day workshop, and especially Claire Gordon, director of the Eden Centre for Education Enhancement at the London School of Economics and Political Science, as the keynote speaker who kicked off our institutional discussion.

Isabel Fischer is professor of digital innovation at Warwick Business School.

If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.