Teaching assistants are usually the last group to get technology training and often the group who'd benefit most. Here's how to train TAs to use AI well in interventions — practically, and with the right boundaries.
Why TAs are the biggest untapped win
TAs run interventions, adapt resources on the fly, and support the pupils with the greatest needs — often with minimal preparation time and no budget. A TA who can produce a differentiated resource in two minutes rather than making do is a genuine gain for the children they support. Yet AI training almost always goes to teachers first, if TAs get it at all.
What TA training should cover
1. The boundaries, first and clearly
TAs are often more anxious than teachers about doing something wrong, and rightly want clarity. Cover the data rules plainly, and be explicit about what they are and aren't authorised to do — see staff acceptable use rules.
2. Adapting resources in the moment
The highest-value skill. A TA who can quickly simplify a text or build a scaffold when a pupil is stuck changes that pupil's lesson.
Prompt — quick adaptation
Simplify this text for a pupil with a reading age of about 8, keeping all the key content and the same level of thinking required: [paste].
3. Making intervention resources
Prompt — intervention activity
Create a 15-minute one-to-one intervention activity for a Year 3 pupil working on [skill, generic]. Include what the adult says, what the pupil does, and a simple way to check whether they've got it. Low preparation, no printing beyond one sheet.
4. Explaining things differently
The core TA skill — finding another way in when the class explanation didn't land.
Prompt — alternative explanations
Explain [concept] in three completely different ways suitable for a Year 4 pupil who didn't understand the classroom explanation. Use a different analogy each time, and flag if any analogy could cause a misconception.
5. Recording and feeding back
Turning rough intervention notes into clear feedback for the teacher — quick, and often skipped for lack of time.
The boundary that matters most for TAs: no pupil names or details go into AI, ever. TAs work with the most sensitive information about the most vulnerable pupils, so this needs saying clearly and more than once — not because TAs are less careful, but because they're given the least training and the most sensitive work.
How to run the training
- Separately from teachers where possible — TA needs differ and mixed sessions default to teacher examples
- Hands-on with their real work, using an actual intervention they run
- On the devices they actually have — no point training on kit they can't access
- Short and repeated, since TA time is fragmented
- With explicit permission to use it — many TAs won't unless told they may
The permission point is the real barrier
In most schools the limiting factor isn't TA capability — it's that nobody has clearly said "yes, you're allowed to use this, here's how, here are the limits." Making that explicit unlocks more than the training itself. See AI for teaching assistants and what CPD staff need first.
Want your TAs trained properly this term?
We run sessions built around what TAs actually do — not teacher examples. AskColin gives schools low-cost monthly support with one-to-one help whenever you need it — so when you're stuck on a real job at 8am, there's someone to ask. Practical, jargon-free, built around your team.
Get one-to-one AI support for your schoolFrequently asked questions
Why should schools train TAs to use AI?
TAs run interventions, adapt resources on the fly and support pupils with the greatest needs, usually with minimal preparation time. A TA who can produce a differentiated resource in two minutes rather than making do is a real gain — yet AI training almost always goes to teachers first.
What should TA AI training cover?
The data boundaries first and clearly, adapting resources in the moment, making intervention activities, explaining concepts in alternative ways, and turning rough notes into feedback for teachers. Hands-on with their real interventions, on the devices they actually have.
What's the most important boundary for TAs using AI?
No pupil names or details go into AI, ever. TAs work with the most sensitive information about the most vulnerable pupils, so this needs stating clearly and repeatedly — not because TAs are less careful, but because they get the least training and the most sensitive work.
What stops TAs using AI in schools?
Usually not capability but permission — nobody has clearly said 'yes, you're allowed to use this, here's how, here are the limits.' Making that explicit often unlocks more than the training itself.