AI’s potential to support exploratory and action research

There can be many benefits for teachers and other professionals in researching their own practice – achieving a better understanding through exploratory and/or action research can improve relationships with learners, form a good basis for appropriate change and enhance teacher agency and motivation (Smith & Rebolledo, 2018). Practitioner research allows teachers to gain new perspectives by developing research questions, gathering classroom data and analysing that data to get deeper insights. Following on from this kind of ‘exploratory research’, a teacher can go further by planning new actions and evaluating their effects, in one or more ‘action research’ cycles.

While many recent initiatives have shown the feasibility and usefulness of teachers engaging in useful classroom research even in relatively large-class, low-resource contexts in Africa, Asia and Latin America, it has become clear that appropriate mentoring is needed to initiate and sustain teachers’ engagement – and spreading the benefits of teacher-research will require more mentors to be trained. Through work with the developers of the award-winning Noticing app, Şirin Soyoz and I have recently been discovering that AI can play a potentially very useful role in supporting such mentoring.

Şirin and I started discussing and experimenting with the capabilities of ChatGPT as a possible teacher-research mentoring assistant in January 2024. Şirin is an experienced teacher-research mentor from Turkiye who has been investigating the capabilities of AI in her work for the British Council. I had also been wondering about the potential of AI to aid with teacher-research mentoring but without the kind of clear focus on actually developing something that Şirin had, and which online interactions with her now gave me. We wrote and tweaked various prompts and inputted information about the issues facing us in our own classroom practice to see what mentoring ChatGPT could give us. However, we soon came up against the limitations both of ChatGPT and of our own capacities to develop a viable tool. ChatGPT was overly helpful, indeed directive, bossy and patronising, giving us all kinds of advice about what we ‘should’ do in our classrooms but not scaffolding reflection or guiding us to formulate our own research questions for further exploration, as we knew a good teacher-research mentor would do.

We therefore decided that we would need to work with a developer, and, having heard recently about ‘Noticing’ and collaborated previously with Elena, I contacted her and Matthew to ask for a trial. We were immediately attracted to the way Noticing offers a non-directive, non-judgmental mentoring experience. This is due – as we subsequently learned from Matthew – to the way several AI platforms are combined to create a particular kind of dialogue partner, Noa, which doesn’t provide all the answers but instead leads the user to deeper reflection and understanding, in an empowering as opposed to deskilling process for teachers. This was impressive: after all, it corresponds with the main reason for engaging teachers to do their own research in the first place.

While Noa is under a continuous process of refinement, it can already mentor a teacher effectively in key areas of teacher-research, for example in formulating research questions. Partly, this has been achieved by inputting authentic mentoring conversations to which we have appended comments in a ‘think-aloud’ kind of way. In the twice-monthly collaborative sessions we’ve been having with Elena and Matthew since March, I’ve found it fascinating to have this opportunity to reflect more deeply and be more explicit than previously about different aspects of the teacher-research mentoring process; for example, when exactly does a teacher-research mentor need to be relatively directive (there are some occasions!)? We’ve also had tantalizing glimpses of how AI could itself be used as a data-gathering instrument for classroom data, so we feel we may even be developing a new form of teacher-research – time will tell!.

We’re now planning to engage in some more systematic collaborative research and piloting which we’ll be reporting on in this blog. Along with practical questions like ‘How can AI be used in a context of person-to-person mentoring?’, there will be some relatively philosophical ones (e.g. ‘What does attempting to develop AI support reveal that we didn’t already know about the mentoring process?). Answers to both kinds of question will help in the further development of AI’s capacities to mentor teacher-research, and it’s clear to me that a degree of usefulness in this area has already been achieved.

For the benefits of teacher-research to be spread more widely, then, more mentors are required, and both mentor-training and practice in mentoring need to be developed. If AI could provide support in this process, acting as a kind of assistant in guiding teachers to formulate viable research questions and to generate, analyse and interpret classroom data, this could help increase the numbers of teachers benefitting worldwide from teacher-research engagement; at the same time, this will provide good opportunities for reflection on effective mentoring for novice and more experienced mentors who use the AI tool we are developing. At least this is our hope, and we will be reflecting here on what we learn as we develop this new app.

Prof. Richard Smith


Acknowledgment: Adapted from a 9 September 2024 blog post on the Noticing Network website.

Image generated using DALL-E, November 2024.

Richard Smith
Richard Smith
Articles: 2

Leave a Reply

Your email address will not be published. Required fields are marked *