In 2025, Jisc launched the AI in Marking and Feedback Pilot, a year-long initiative bringing together colleges and universities to explore whether AI can meaningfully reduce marking and feedback workload in a way that is acceptable to key stakeholders.
The project spans two strands — tools designed specifically for educational purposes (these were Graide, Keath and TeacherMatic), and general-purpose AI tools (such as ChatGPT, Gemini, and Copilot), where the project focuses on custom assistants.
Across the initial months of the pilot, insights have been collected via regular community sessions, feedback forms and one-to-one interactions. As common themes appear, we’d like to share insights with the wider Jisc membership, allowing learnings to translate to direct value for the sector.
As such, we’re publishing this series of blogs, which we hope will give you a useful window into what the pilot has revealed about the role of AI within marking and feedback. You can read all the blogs in this series by following the links on this parent page.
Ethics, in Principle and Practice
The pilot cohort recently took part in a series of sessions exploring the ethical use of AI in marking and feedback. The first of these sessions focused on the principles of using AI ethically in marking and feedback, the second on the practices and processes needed to realise such underlying values.
During the initial session, three sets of ideals stood out:
- respect for privacy and ownership
- awareness of broader social and environmental impacts
- a commitment to fairness
Within the second session, the favoured mechanisms for achieving ethical AI in practice were:
- keeping humans in the loop
- transparency for students
- staff training and confidence-building
- having a clear and specific rationale for each particular use of AI
Below we unpack some of the key ideas that were raised during these conversations.
The principles
Respect for Privacy and Ownership
Participants stressed that ethical use begins with respecting students’ rights over their work. This includes clarity about how student submissions are processed, what data AI tools retain, and how underlying models operate.
Many also highlighted that transparency is not merely declarative: it requires staff to have the capability and confidence to understand how tools handle data and to raise questions when something is unclear.
Attention to Social and Environmental Impacts
Ethical use was also framed in terms of wider consequences beyond the immediate marking task. Participants reflected on the environmental footprint of unnecessarily drawn-out conversations with AI tools (i.e. involving lots of backing-and-forthing, perhaps symptomatic of lower confidence with prompting), and the potential long‑term shifts in academic work and professional identity as automation becomes more commonplace.
Although views varied, a shared principle emerged: to use AI ethically is to be mindful of the wider consequences of doing so. In particular, institutions should try to minimise unnecessary environmental impact and think carefully about how people can be empowered by AI, rather than being marginalised by it.
Commitment to Fairness
Fairness was a central principle running through the discussions. One noteworthy insight here was the risk of unequal experiences across an institution, if AI‑assisted approaches are adopted unevenly or without clear, centralised guidance.
The practices
Human in the Loop
The strongest consensus across the session was clear: AI has a role, but professional expertise remains essential.
Participants emphasised that policies should be in place to ensure human expertise and judgement remains central to the marking and feedback process. Educators, moreover, must remain ultimately responsible and accountable.
Alongside policies mandating the human-in-the-loop approach, support should be provided to ensure this can be achieved in practice. On this point, Jisc’s AI team has recently published a blog exploring the issue of how educator oversight can be maintained.
Transparency for Students
Another strong theme was the importance of openness with students around staff use of AI in marking and feedback.
The cohort noted that institutions should be explicit about where, how, and why AI tools are applied in marking processes.
A traffic‑light‑style disclosure system—mirroring those already used for student AI use—was suggested by a participant at Swansea University as a way to show what role AI played in staff marking. Green, for instance, could mean that no AI was used in the marking process. Amber: that AI was used, but the feedback was reviewed and edited extensively by an educator before being returned. While red could indicate that a human marker had quickly reviewed a piece of work and the feedback to ensure reasonable alignment (suitable for first drafts or other such formative pieces of work, perhaps). Alternatively, red could be reserved for human-free marking, which might only be suitable in a limited range of contexts.
Staff Training
Another widely supported mechanism was the need for structured staff training covering both the opportunities and the limitations of AI in marking.
Participants highlighted that training should help staff understand:
- when AI tools are appropriate to use
- where AI struggles—especially in higher-level, interpretative, or judgement-heavy assignments.
- how to interpret AI outputs critically
This directly supports a human-in-the-loop model: staff need the skills to question, validate, and contextualise AI generated-content rather than treating its outputs as beyond reproach.
A Clear Rationale for Using AI (Not Every Assessment Is the Right Fit)
Bringing together the themes of staff skillsets and transparency with students, the cohort suggested that before a particular use of AI within the marking and feedback process is signed off, a rationale document should need to be completed.
Such a document should highlight:
- why AI is being used
- what problems and being solved/opportunities seized
- the benefits to different stakeholder groups
- the risks and how these are being managed.
Concluding thoughts
Across these two sessions, a clear pattern emerges: ethical principles only become meaningful when they are translated into concrete practices.
The cohort’s discussions suggest that the principles identified — privacy and ownership, fairness, and awareness of wider impacts — do not operate in isolation. Instead, they place specific demands on how AI is implemented in practice. Staff training enables educators to interrogate AI outputs critically; clear rationales ensure that AI is used deliberately rather than by default; and transparent communication helps maintain trust with students.
As the pilot continues, this relationship between principle and practice is likely to remain central. Further work will be needed to support institutions in turning high-level ethical commitments into consistent, scalable approaches that can be applied across different contexts.
Find out more by visiting our Artificial Intelligence page to explore publications and resources, learn more about our communities and sign up for our AI Literacy training.
For regular updates from the team sign up to our mailing list.
Get in touch with the team directly at AI@jisc.ac.uk
