AI assessment tools are no longer something schools can treat as a distant possibility.
They are already being explored and used to support writing assessment, feedback, analysis and teacher decision-making. For school and MAT leaders, that creates real opportunities, but it also raises some important questions.
I recently joined Assessment in the Age of AI: What School and MAT Leaders Need to Get Right, a Pobble webinar chaired by Al Kingsley MBE, alongside Hannah Gillott from Stylus and Chris Toyne from Air. What I really valued about the discussion was that it did not become a conversation about whether AI is simply “good” or “bad”.
Instead, we kept coming back to something much more useful:
How do we make sure AI genuinely supports assessment, while keeping teacher judgement, professional dialogue and shared understanding at the centre?
These are the seven questions I think school and MAT leaders should be asking.
This was the first point I made during the discussion, because I think it has to come before anything else.
Before choosing an AI tool, we need to be clear about the problem we want it to solve.
Is it workload?
Consistency?
Formative insight...into a class, a year group, or across a whole trust?
Support for teachers when identifying patterns in writing?
There are lots of perfectly valid reasons to explore AI, but they may lead to very different decisions.
For me, the important thing is to keep coming back to that original question:
What problem are we trying to solve?
It is very easy to start with the technology and work backwards.
A much stronger approach is to identify the need first, then decide whether AI is actually the right way to address it. Or indeed whether AI alone will solve it.
Hannah raised a really useful idea during the discussion around non-negotiable human thinking.
AI can support more and more parts of the assessment process, but that does not mean schools should automatically hand those things over.
Leaders need to ask:
What do we still need teachers to notice?
What do they need to understand?
What professional thinking do we want them to practise?
What decisions must remain theirs?
In writing assessment, that might include understanding the overall quality of a piece, considering the writer’s intent, recognising whether skills have been applied independently, and making the final assessment judgement.
The question should not simply be, “What can AI do?”
It should be, “What do we still need our teachers to do, understand and own?”
Chris raised another question that I think schools need to ask honestly: Are we ready? AI adoption is often treated as a technical decision, but in reality it is just as much about people, culture and professional development. Introducing a tool successfully requires more than appointing an “AI lead” and hoping they can carry the responsibility alone.
Schools need to think about:
Al Kingsley made the point that schools are used to having established frameworks, training and professional communities around many areas of practice.
AI is moving much faster. That makes it even more important to build understanding across the organisation rather than pinning responsibility on one person.
One thing we all agreed on was that AI can offer real benefits.
It can process large amounts of information quickly.
It can surface patterns across a body of writing.
It can help teachers identify strengths, gaps and areas worth investigating. Perhaps some that, as a teacher, we had not yet even fully realised ourselves.
Hannah shared a great example of how it can retain and compare information over time in a way that is difficult for any individual teacher managing a full class or multiple classes. Teachers may remember the pupils they have spoken to in depth about a particular writing issue, but it is almost impossible to retain every detail about every pupil and every skill over several months.
It can help surface patterns again when they become useful or when further evidence is gathered and added at a later date.
For me, that is where the value lies: it can give teachers a richer starting point for professional thinking.
But it is still a starting point. Not the final judgement.
5. What can AI not do better?
This became one of the clearest themes of the whole conversation.
A tool can analyse what is on the page, but we all know that writing assessment depends on understanding so much more than just analysing a finished text.
Only the teacher may know:
how independently the writing was produced
what scaffolding or modelling was used
what the pupil was trying to achieve
whether a particular choice was deliberate
how securely a skill has been demonstrated over time
the wider context around a child's learning and outcomes
I kept coming back during the discussion to independence and authorial intent.
A piece of writing may look impressive in isolation, but the teacher knows the context in which it was produced. This is really crucial. Chris made a similar point. AI can surface evidence and stimulate discussion, but it cannot fully account for all the context teachers bring to a judgement. That is why writing assessment cannot simply become a case of accepting whatever a tool outputs.
6. What happens when the teacher and the AI disagree?
This is a very important discussion to have. Hannah spoke very openly about the fact that AI and teacher assessments will not always agree. Neither will two teachers. The important thing is what happens next. A disagreement should not automatically mean the AI is wrong. Equally, the tool’s output should never be treated as a verdict that the teacher simply accepts. Instead, disagreement can become the starting point for professional discussion.
Why has the tool reached that conclusion?
What evidence is it highlighting?
What does the teacher know that the tool does not?
What does the wider body of writing show?
Does the judgement still feel secure?
During the webinar, I talked about how moderation conversations could increasingly begin with the pieces or judgements teachers are least sure about, rather than the ones they feel most confident discussing. That is where some of the richest professional learning can happen.
AI may help surface uncertainty, and then effective and supportive moderation discussions give teachers the space to explore it.
7. If AI becomes part of assessment, does moderation matter more or less?
The answer to this was clear to all of us: AI does not remove the need for moderation. If anything, dedicated time for professional discussion becomes even more important. Moderation is not simply about checking whether a judgement is right. Effective moderation is where teachers share their thinking, challenge assumptions, build shared understanding, develop and develop their own assessment expertise. Time for this is so important if you want to ensure teacher expertise is continually developed.
Might the focus within a moderation change as schools begin using AI-supported assessment tools? Yes possibly.
Might teachers arrive with more information? Might they have clearer patterns to explore within the conversation?
They may have questions raised by an AI analysis that they come to the meeting ready to want to explore. But the professional conversation still needs to happen. And the teacher still needs to own the final judgement.
So what should leaders do next?
One of the strongest themes running through the webinar was that successful AI adoption is not really about the technology on its own. It is about how the technology fits into the wider (human-owned) assessment process.
That means asking:
Why are we using it?
What will teachers still own?
How will we know whether it is helping?
What professional development will staff need?
How will we handle disagreement?
Where will moderation fit?
And how will we make sure teachers remain confident enough to explain and stand behind the judgements they make?
Used effectively, AI can reduce some of the mechanical work, surface useful evidence and help teachers see patterns they might otherwise miss. That is ever so valuable. But assessment still depends on human understanding. Teachers bring the context. Teachers know the pupil. And teachers make the final judgement.
Where Pobble fits
This is exactly where I see Pobble’s role. Pobble supports schools and MATs with expert, human-led writing moderation. Our role is not to validate AI tools or their outputs. It is to help schools strengthen the teacher assessment judgements being made around them.
Through professional dialogue, training, strategy support and moderation sessions, we help teachers understand standards and make robust decisions. We help leaders build greater confidence and consistency across schools. Ultimately, ensuring your moderation sessions are positive and purposeful.
AI can support the assessment process. Pobble supports the professional judgement around it. And as AI becomes a bigger part of writing assessment, I think that combination is likely to matter more, not less.
You can watch Assessment in the Age of AI: What School and MAT Leaders Need to Get Right to hear the full conversation with Al Kingsley MBE, Hannah Gillott, Chris Toyne and me.
Need help getting the balance right?
If your school or trust is already using AI-supported assessment, or you are considering how it could fit into your approach, I’d be very happy to help you work through the practical questions.
Book a strategy call with Laura
Or explore Pobble Moderation Services to find out more about external validation, training, moderation support and statutory assessment services.
Explore Pobble Moderation Services
Laura Bailey
Head of Moderation and Assessment, Pobble
Connect with Laura on LinkedIn.
Pobble Education Ltd,
Rosehay,
Tremorvah Wood Lane,
Truro, TR1 1PZ,
Cornwall, UK