When Stephen Morgan, minister for early years, announced last week that he is going to spend £4 million on creating AI tools to help teachers to plan and feed back to students, it would have been fair to expect a cheer in response. But I don’t think I’m alone in feeling some disquiet.
I have no doubt that the intentions are good. Workload is a frequently cited reason for teacher dissatisfaction and freeing up teachers to give more time to pupils in the classroom has got to be a good thing. So why are some of us worried?
The problem is that this approach misunderstands what teachers do when they are planning.
When I’m preparing to teach a unit on states of matter to Year 4, I’m not just arranging concepts from the set of knowledge I need to teach. I’m checking my own comprehension. I’m thinking about likely misconceptions, mine as well as the pupils. I’m wondering how I’ll keep Billy on board given what has been going on for him over the summer. I’m wondering how to deal with lockdown gaps in children’s knowledge.
The planning is as much about me and about my pupils as it is about the material. This is true whether you are in a school where each unit has a pre-existing booklet or in a school using very different methods.
It’s still about me, the teacher, figuring out how to get these particular children to learn this stuff. It is deeply human and it is at the heart of what makes teaching a profession and a craft.
The same goes for the intention to assist with feedback. Teachers know that the most effective feedback happens in the moment, at the point of learning. A thousand tiny interactions and redirections. It is not a sticker, printed and stuck in an exercise book, or an auto-generated comment on a Google Doc.
Feedback is a continuous process in lessons and is as much to inform me as it is to teach the child. What have they understood? Why haven’t they got this? How can I reframe the question?
‘Different tasks for different students’
The government document suggests using AI to create individual resources for children with particular needs. This contradicts the current movement in the profession away from differentiated tasks and towards scaffolding such that all can achieve.
Again, this is no doubt well-meaning but risks creating many different tasks for different students, which are time-consuming and distracting to assign and monitor.
Some aspects of the project do deserve applause: teachers already using freely available AI tools such as ChatGPT will have found that they get some very dubious responses due to the unreliable quality and provenance of the material that the AI model is trawling for results.
The government wants to spend the first £3 million to create a curated pool of quality materials that AI developers can draw on, including exemplar planning and samples of pupil work. Not a bad idea at all (although we might raise an eyebrow and ask who defines what a good lesson plan looks like).
Colleagues who have been in the profession for a while will remember the list of tasks that teachers should not be expected to do: bulk photocopying, filing, data entry and more.
That list was predicated on tasks that did not require the teacher’s professional judgement. If there is anything that requires the professional judgement of the teacher more than what to teach, how to teach it and how to communicate with pupils, I don’t know what it is.
Ed Finch is one of the founders of #BrewEd and tweets at @MrEFinch
For the latest research, pedagogy and practical classroom advice delivered directly to your inbox every week, sign up to our Teaching Essentials newsletter
