AI for Teacher Training: Developing Great Teachers | TeacherMatic

Watch the webinar: Supporting Trainee Teachers with AI

Watch the 54-minute session exploring AI-supported observation, feedback, mentoring, coaching and professional development, alongside the Newham College case study.

Watch the webinar on YouTube

Why trainee teachers need scaffolding, not simply access to AI

A generic generative AI tool can be extraordinarily useful. But a blank prompt box carries an important assumption: that the person using it already knows what to ask.

An experienced teacher may know that an effective lesson plan should connect learning objectives, activities, assessment, differentiation and learner needs. They may know how to challenge a weak AI response or identify when an activity is educationally inappropriate.

A trainee teacher may still be developing that knowledge.

That is why one of the principles we use when discussing TeacherMatic is:

“The interface is the pedagogy.”

Rather than beginning with a blank box, TeacherMatic’s educator-focused generators guide users through structured questions.

What are the learning objectives? Who are the learners? What needs should be considered? How should learning be assessed? Where might differentiation or SEND support be required?

The purpose is not simply to produce an output more quickly. The structure itself helps model some of the questions an experienced educator would consider.

A trainee who has never constructed a rubric, for example, can see what information is required to build one. A new teacher planning an activity is prompted to consider the learners and intended outcomes rather than beginning only with a topic.

This makes AI potentially valuable not simply as a productivity tool, but as a form of pedagogical scaffolding.

Supporting the whole teacher-development cycle

During the session, we explored TeacherMatic through a simple teacher-development cycle:

The important point is that these activities should not operate independently.

An observation should lead to thoughtful feedback. Feedback should inform mentoring and reflection. Reflection can lead into coaching. Coaching can shape professional development targets. Those targets then provide a focus for the next observation.

AI can help with the preparation and structure at each point — while the teacher educator or mentor remains responsible for the professional judgement.

1. Preparing for observation

Teacher educators are frequently asked to observe lessons beyond their own subject specialism.

Someone from an IT or English background might, for example, be asked to observe a trainee delivering a practical welding session.

The Lesson Observation Prompts Generator can help the observer prepare appropriate “look-fors” before entering the room, combining relevant teaching and learning considerations with the context of the subject.

The value here is not that AI decides what good teaching looks like.

Instead, it can help the observer prepare better questions and identify areas worth paying attention to. The mentor can then select, amend or reject those suggestions based on the purpose of the observation and the needs of the trainee.

2. Turning observation notes into developmental feedback

Observation itself can be only one part of the workload.

The write-up that follows often takes considerable time, particularly when an observer is balancing teaching with mentoring or quality responsibilities.

The Lesson Observation Feedback Generator can use notes from an observation — including typed or appropriately captured handwritten notes — to help structure a professional report.

But there is an important distinction:

AI did not observe the lesson. The educator did.

The generated feedback therefore needs to be checked against what actually happened.

Does the wording accurately represent the evidence? Has the AI inferred something that was not observed? Are the priorities right? Is the tone developmental? What should the trainee realistically work on next?

The human review is not an optional final step. It is the most important part of the process.

3. Moving from feedback to reflection

Observation feedback becomes far more useful when it leads into professional thinking.

The Mentoring Prompts Generator, which draws on Egan’s Skilled Helper approach, can help create questions based on a development theme or the evidence from an observation.

Rather than telling a trainee what to do, the mentor can select questions that encourage them to examine their own practice.

What happened, and why might it have happened? What impact did it have on learners? What alternatives could be considered, and what will you try next?

AI may help generate the questions. The quality of the mentoring conversation still depends on the mentor.

4. Supporting coaching conversations

Coaching takes this principle further.

Using models such as GROW, TeacherMatic’s Coaching Prompts Generator can provide a structured starting point for a developmental conversation.

This can be particularly useful for mentors who understand teaching well but are less confident about structuring a coaching session.

The intention is not for AI to become the coach.

Good coaching relies on listening, trust, professional sensitivity, context and the ability to respond to the person in front of you.

The generator provides preparation. The human provides the relationship.

5. Turning reflection into action

Development also needs to result in something practical.

The CPD Action Planner can use identified areas for development to help create structured, achievable actions and SMART targets.

The same evidence gathered during an observation can therefore contribute to feedback, mentoring, coaching and professional development planning.

That creates a more connected development journey:

Generate → Review → Reflect → Refine → Apply professional judgement → Use

From paperwork to professional dialogue

Throughout the webinar, we used an illustrative 80/20 approach. This is a way of thinking about the division of work, rather than a scientific formula or a guaranteed saving.

It is simply a helpful way of thinking about the relationship between an educator and AI: perhaps AI can complete a significant proportion of the initial groundwork, but the remaining human contribution — reviewing, reflecting, contextualising and refining — is where much of the professional value sits.

That matters because workload reduction is not the end goal.

The question is what educators can do with the capacity they regain.

If an observation report takes less time to structure, could more time be spent discussing the lesson with the trainee?

If planning resources can be generated more efficiently, could the trainee spend longer thinking about why an activity is suitable for those particular learners?

If mentoring administration becomes easier, can the mentor focus more on listening and asking better questions?

The most valuable outcome of AI may not be producing more things.

It may be creating more time for the human work around them.

What this looks like in practice: Newham College

Shirley Green, Teacher Training and Development Manager at Newham College of Further Education, provided one of the most important perspectives during the session: what happens when AI is embedded into teacher development rather than simply made available as another piece of software.

Newham College first piloted TeacherMatic in 2024 as part of the Local Skills Improvement Plan programme.

Shirley saw particular potential for trainee teachers.

New teachers were already dealing with a considerable cognitive load: pedagogy, learner needs, assessment, lesson planning, differentiation, learning theories and classroom practice all had to be developed simultaneously.

Some trainees lacked the confidence or experience to design effective lessons independently, while significant time was being spent creating resources and planning activities.

The response at Newham College was not simply to issue licences and expect trainees to work things out for themselves.

TeacherMatic was integrated into the teacher-training experience.

During group tutorials, trainees could discuss what they were teaching that week and then use an appropriate generator to create something they genuinely needed for the classroom.

A first-year trainee might leave with a starter activity ready to adapt and use the following day. A more experienced trainee might build a starter, main activity and exit task.

The crucial next step was always review.

Trainees were encouraged not to simply accept an AI-generated resource but to ask:

• Is this appropriate for my learners?
• What learning does this actually support?
• Is it fit for purpose?
• What needs adapting?
• How could I improve it?

That process changes the nature of AI use.

Instead of replacing thinking, the output becomes something to think about.

Making AI a stimulus for professional thinking

Newham College also connected AI use with peer discussion and coaching.

In its unseen observation model, trainees could prepare a lesson and discuss the planning and rationale with a peer beforehand. They would then teach the lesson without being formally observed and return afterwards to discuss what happened.

The peer could take on a coaching role.

Did the resource work as expected? Were learners engaged? What would you change next time, and why?

TeacherMatic therefore became part of a larger reflective process rather than the end of the process.

As Shirley put it during the session, used in this way TeacherMatic can become a “stimulus for professional thinking.”

Newham College also created a TeacherMatic forum where trainees could share resources, ideas, experiences and effective uses of different generators. This helped keep AI use visible and connected to continuing professional dialogue rather than becoming a one-off training activity.

For organisations introducing AI, this is an important lesson: access needs to be accompanied by practice, discussion, modelling and reflection.

What did Newham College report?

Newham College’s own evaluation offered useful indications of impact.

Trainees initially reported saving around two to three hours per week, with savings reaching up to five to six hours per week as they became more familiar with the tools and with reviewing and refining the outputs.

Newham College also reported that:

• 90% reported improved creativity in planning.
• 68% reported higher student engagement.
• Trainees demonstrated gains in areas including planning for learning and assessment for learning.
• 75% of active users reported using TeacherMatic daily or weekly.

These are Newham College’s own reported findings rather than claims that every institution should expect identical results.

The deeper point is perhaps more important than the percentages.

Shirley described seeing richer planning: more varied activities, clearer assessment opportunities, differentiation and greater attention to inclusion.

Saving time mattered because it enabled trainees to produce and explore more possibilities — and then discuss and evaluate them.

Building AI-literate teachers

Initial teacher education also has a wider responsibility.

New teachers are entering a profession in which generative AI will increasingly form part of the educational landscape.

Teacher development therefore cannot focus only on whether a trainee can use an AI tool.

They also need to learn when to use it, when not to use it, how to question it, how to improve it and when their own professional judgement must take over. This means developing the ability to use AI critically within professional practice.

A trainee teacher who accepts every generated lesson plan without question has learned very little.

A trainee who can explain why they rejected an activity, changed the level of questioning, adapted a task for a particular learner or challenged an AI-generated assumption is developing professional judgement.

That is the kind of AI literacy teacher-training programmes should be encouraging.

Responsible AI and professional judgement

Responsible AI use also needs to be explicitly taught rather than assumed.

The TeacherMatic SAFE AI Framework centres on four principles:

Explore the TeacherMatic SAFE AI Framework

A useful principle from the session summarises this well:

AI can support the professional conversation. It should not become the professional conversation.

Co-designing the next generation of ITT tools

The webinar also provided a first look at new TeacherMatic tools being developed specifically for Initial Teacher Education.

Richard Goddard demonstrated a beta ITT mentoring support tool, shown in the captured platform interface as Teacher Training Support, designed to bring together information such as mentoring discussion notes or transcripts, observation evidence, relevant teaching standards, potential actions and feedback for the trainee.

At the September webinar, it was presented as a tool still in development. TeacherMatic’s approach is based on co-design with educators, with testing and feedback helping to shape the next version.

Teacher trainers, mentors and institutions are encouraged to test tools, identify what works, highlight what does not and suggest what would make them more useful in authentic teacher-training contexts.

Feedback then informs subsequent development.

For educational AI, this kind of co-design matters.

A technically impressive tool is not necessarily a pedagogically useful one.

The people who understand the workflow, pressures, language and professional responsibilities of teacher education need to help shape the technology intended to support them.

What can ITT providers do next?

ITT providers can begin with a focused activity, supported by their institutional expectations for responsible AI use.

A few practical starting points emerged strongly from the session.

Introduce responsible AI as part of induction. New teachers should discuss appropriate AI use, privacy, professional judgement and critical evaluation from the beginning of their training.

Embed AI into authentic teacher-development activity. Move beyond demonstrations. Use it with an actual lesson the trainee needs to plan, an observation that needs to be written up or a mentoring conversation that is genuinely taking place.

Make critique part of the task. Do not only ask trainees to generate something. Ask them what they changed, what they rejected and why.

Connect AI with mentoring and coaching. Generated outputs become much more valuable when they form the basis of professional dialogue.

Create opportunities to share practice. Forums, communities of practice and peer-led sessions help trainees see different approaches and discuss what effective AI use looks like.

Keep professional judgement visible. Reviewing and refining should not be treated as correcting the AI afterwards. They are fundamental parts of the professional process.

Giving teachers time back without taking teachers out of teaching

There is a temptation to judge educational AI by how quickly it can create something.

For teacher education, that is too narrow.

The more important question is whether it helps us develop more thoughtful, reflective and capable teachers.

Does it reduce unnecessary workload and scaffold a trainee who is still learning what good planning looks like? Does it help a mentor prepare a better conversation and create more time for professional dialogue? Does it encourage trainees to question, adapt and justify their decisions?

Used carefully, AI can support all of these things.

But the teacher, mentor and teacher educator must remain at the centre.

As Shirley concluded during the webinar, TeacherMatic should not simply be viewed as another piece of technology. It can form part of a wider approach to developing AI-literate teachers who understand both how to use AI and when to rely on their own professional judgement.

Perhaps the most important ambition is also the simplest: AI should give teachers time back without taking teachers out of teaching.

Continue exploring TeacherMatic

Explore practical support for teacher training, continue your professional learning and keep responsible AI use at the centre of your practice.