The session explored one increasingly urgent question facing the sector:

How can institutions use AI in assessment and feedback responsibly, transparently and with meaningful human oversight?

Rather than framing AI as a replacement for educators, the webinar focused on something far more important: how institutions can develop ethical, sustainable and pedagogy-first approaches that support educators while maintaining trust, fairness and professional judgement.

AI Is Already Here. Policy Often Isn’t.

One of the strongest messages from the session was the need for honesty about where education currently stands.

AI tools are already being used across institutions, often informally and without consistent guidance. Students are experimenting independently, while staff attitudes towards AI-supported workflows vary significantly across departments and organisations.

This creates what the SAFE Framework presentation described as a growing clarity gap.

• Some staff are using AI extensively.
• Others are avoiding it completely.
• Students receive inconsistent experiences.
• Quality teams remain uncertain.
• Policies are still evolving.
  
  

The result is inconsistency, anxiety and unmanaged risk. The challenge is no longer whether AI will be present in assessment workflows. The challenge is whether institutions can establish clear guardrails around its use before informal practices become embedded without oversight.

The Fear Many Educators Share

A major theme throughout the discussion was the concern around “AI marking AI”.

Students using AI to generate assignments, followed by AI being used to assess those same assignments, with little meaningful human involvement in between, represents a scenario many educators are understandably uncomfortable with.

Importantly, the webinar did not dismiss these concerns. Instead, it treated them as legitimate and necessary starting points for discussion.

Questions around academic integrity, authenticity, fairness and reliability are not barriers to innovation. They are exactly the questions institutions should be asking before implementing AI-supported assessment workflows at scale.

This human-in-the-loop model became the central thread connecting the entire webinar. AI may support drafting, structure feedback or assist moderation processes, but final academic judgement remains with qualified educators.

That distinction matters enormously.

Why Institutions Are Exploring AI-Supported Feedback

Alongside concerns, the webinar also acknowledged the growing pressures surrounding assessment workload across schools, colleges and universities.

Large class sizes, increasing expectations around feedback quality, recruitment challenges and staff wellbeing pressures are creating unsustainable strain across the sector.

The session openly addressed realities many educators recognise immediately.

• Evenings and weekends spent marking.
• Fatigue affecting consistency.
• Delayed feedback cycles.
• Increasing pressure to personalise responses at scale.
  
  

Within this context, AI-supported workflows were explored not as a replacement for educators, but as a potential mechanism for reducing administrative burden while improving consistency and turnaround times.

Potential benefits discussed during the webinar included faster feedback delivery, more structured and detailed feedback, improved consistency, reduced fatigue during marking, support for moderation and standardisation, and potential reduction in unconscious bias.

Crucially, these benefits were always framed within the context of professional oversight and institutional governance.

Demonstrating AI-Assisted Feedback in Practice

The session included demonstrations of TeacherMatic’s Advanced Feedback Generator and related assessment workflows.

Attendees were shown how educators can upload assignment briefs and marking rubrics, generate structured draft feedback, review AI-generated grading suggestions, annotate student submissions, add and refine human comments, download annotated feedback documents, and compare submissions for moderation and standardisation purposes.

Particular attention was given to consistency. The webinar demonstrated how repeated uploads of the same student work produced highly similar grading outcomes while still generating nuanced variation in feedback commentary.

This led to wider discussion around how AI-supported workflows could potentially support moderation, standardisation, quality assurance, staff calibration and feedback consistency across teams.

The session also explored specialist BTEC-style workflows aligned to Pass, Merit and Distinction criteria.

Introducing the SAFE Framework

The second half of the webinar introduced the SAFE Framework, presented by Esam Baboukhan.

The framework was positioned as a practical starting point for institutions seeking to approach AI-supported feedback and assessment responsibly.

Rather than presenting SAFE as a rigid compliance framework, the session positioned it as a shared language institutions can use to navigate increasingly complex questions around AI governance, assessment integrity and institutional trust.

Governance Before Scale

Another strong theme throughout the webinar was the importance of staged implementation.

Rather than encouraging institutions to deploy AI-supported assessment at scale immediately, the session emphasised careful pilots, structured evaluation and clear governance processes.

Suggested implementation considerations included:

• Establishing cross-functional implementation groups.
• Running structured pilots.
• Reviewing institutional policies.
• Consulting awarding body guidance.
• Providing staff CPD and training.
• Involving students in the conversation.
• Creating transparent appeals procedures.
• Monitoring impact over time.

The emphasis throughout was not on moving slowly, but on moving responsibly.

The webinar consistently reinforced that governance, transparency and culture are just as important as the technology itself.

This Is Ultimately About Trust

Perhaps the most important takeaway from the webinar was that AI in assessment is not simply a technical conversation.

It is a conversation about trust.

• Trust in professional judgement.
• Trust in institutional processes.
• Trust in fairness and transparency.
• Trust that students are being assessed responsibly.
  
  

The webinar argued that avoiding AI entirely is unlikely to be sustainable. But equally, adopting AI without clear principles, oversight and accountability creates significant risks for institutions, educators and learners alike.

The challenge now is ensuring innovation does not outpace governance, and ensuring educators remain firmly at the centre of assessment and feedback, exactly where they belong.