AI-Assisted Teacher Reflection: Thinking Before, During, and After the Lesson

Many teachers will likely describe teaching as a profession of doing – and that’s not a wrong perspective. But for me,¹ teaching is also a profession that, in its core, requires thinking. In its most obvious sense, this is thinking about what to teach and how; but it also involves thinking about what is happening in the room, what worked, and what to do differently next time. This is what has been called reflective practice, which –I would argue²— is one of the things that distinguishes purposeful professional action from merely competent work. That much is uncontroversial, I think. The interesting question we are now facing is: what does reflective practice mean in an age when thinking is often outsourced to, or at least assisted by, artificial intelligence?

In this post, I draw on our ongoing work in the Artificial Intelligence for Language Education project to discuss three types of teacher reflection and how AI tools can support each of them. The tri-partite framework around which this post is structured (reflection for, in and on action) comes from the work of Thomas Farrell, who distinguishes between reflection that happens before a lesson, during it, and after it.³ I argue that each type creates a different kind of thinking space, and each offers different possibilities for AI to play a useful (though carefully managed) role.

Before the lesson:
Reflection-for-action

Reflection-for-action is the thinking a teacher does before teaching. Such thinking includes planning, anticipating problems, making decisions about materials and sequencing, and perhaps even asking oneself whether what one is about to do actually serves the learners’ needs. For most teachers, such preparation or lesson planning is routine action, though often carried out under time pressure and perhaps without much opportunity to think things through systematically.

When used responsibly, artificial intelligence can be a surprisingly useful thinking partner at this stage. To be clear, this does not involve offloading planning to AI: the tool does not know our learners, and even if it knows our curricular demands, priorities and preferences, it cannot know our day-to-day disposition. But despite such limitations, it can help a teacher stress-test her ideas, generate alternatives, and make her pedagogical reasoning more explicit.

Reflection for action has a technical and an intellectual aspect. The technical aspect is mostly about preparing materials and instructional sequences. The intellectual aspect is about consciously bringing one’s values, knowledge of the class, and professional experience to bear on pedagogical decisions about a future lesson. AI can accelerate parts of that process by taking the lead in the technical aspects (e.g., proposing activity sequences, drafting differentiated materials, or formatting a lesson plan), freeing up space for the intellectual aspect. Such a division of labour ensures that the thinking that makes those decisions purposeful is still the teacher’s.

During the lesson:
Reflection-in-action

Reflection-in-action is harder to describe because it happens fast — so fast, in fact, that it sometimes doesn’t even register as reflection. Donald Schön, who first described this kind of thinking, called it “thinking on your feet.”² It is the rapid, often tacit decision-making that teachers engage in as a lesson unfolds: noticing what is happening, reading the room, and adjusting without stopping the flow of the lesson. Edd Asp-Miyanishi, whose doctoral work was one of the most interesting theses I examined, described it as the Skilled Teacher Approach: the ability to identify affordances for action and adjust teaching to them.⁴ In the Intentional Dynamics of TESOL, Juup Stelma and I related it to contingent professional action, the semi-conscious adjustment to the dynamic professional landscape.

When AI tools are involved in a lesson, reflection-in-action becomes both more important and more demanding. On one hand, artificial intelligence frees up some cognitive space: for instance, if a chatbot is managing a conversation task, the teacher can circulate, observe, and intervene more deliberately. On the other hand, there are new things to notice and respond to, such as the quality of AI-generated output, whether learners are engaging critically or copying responses, and whether the AI is introducing language that is inappropriate for the level or context.

This is what reflection-in-action looks like: a rapid judgement call that requires the teacher to weigh the AI’s output against their knowledge of the learners, the context, and the communicative goals of the lesson. There is no algorithm for this. It is precisely the kind of professional thinking that AI cannot replicate. I would argue that this becomes more, not less, important as AI tools become more present in classrooms.

After the lesson:
Reflection-on-action

Reflection-on-action is the kind of thinking that takes place after the lesson, when the immediate pressures of teaching have receded. This is slower and more analytical thinking, and some of us⁵ may have been asked to engage with it in formal ways during our teacher education – e.g., in reflective journals. The value of this kind of thinking is that it helps us to connect individual classroom events to broader understandings of teaching and learning, which is how professional knowledge develops over time.

AI can support this phase in several practical ways. If we have a record of a lesson, perhaps in the form of notes taken during or immediately after it, AI tools can recreate a narrative reconstruction of the lesson that makes it easier to revisit what actually happened rather than relying on memory.

Artificial intelligence can even assume the role of “critical colleague”. To do this, a teacher could prompt a conversational agent to ask probing questions about a lesson, suggest alternative interpretations, or draw attention to patterns that might otherwise go unnoticed. A simple prompt like the following can open up a useful reflective conversation:

Assume the role of a critical friend and experienced language educator. Help me reflect on today’s lesson. Ask probing questions about what worked well, what I might do differently, and how the AI tools I used influenced the dynamics of the class.

The key word there is probing. The value of this kind of AI-assisted reflection is not that the AI provides answers, but that it asks questions one might not have thought to ask oneself. This brings to mind a quote by one of my university teachers, Julian Edge, who pointed out that “We learn by speaking, by working to put our thoughts together so that someone else can understand them” (Edge, 2002, p. 19).⁶

Used well, artificial intelligence can push reflection beyond the comfortable and the habitual, which is, after all, what the best critical colleagues do.

A note on what AI cannot do

Running through all three phases is a point that I believe we should keep in mind. Artificial intelligence can generate materials, ask questions, produce transcripts, and suggest alternative ways of doing what we do.

What it cannot (yet?) do is know our learners, share our values, or carry the professional responsibility that comes with teaching. Reflection is valuable precisely because it connects teaching decisions to the human judgement of the teacher making them. AI can extend and support that judgement; it cannot replace it.

This reminder for caution is not a reason to be sceptical about using AI for reflection. It is a reason to be intentional: to use it in ways that sharpen your thinking rather than substitute for it.

Where to go next

The ideas in this post are developed in more depth in the AI Lang Moodle platform, which includes a dedicated module on teacher reflection in AI-assisted language education. The module — which we are still refining with help from your valuable feedback — covers each of the three phases described here with worked examples, diagrams, and tasks designed to support your own reflective practice. It is freely accessible, and we’d love to see you there.

---

Footnotes

1. And perhaps also Henry Giroux, whose 1988 book, Teachers as intellectuals, defined teaching as intellectual work. ↩︎
2. Along with people like John Dewey, D. A. Schön, and –in TESOL– T.S.C. Farrell. You can read more about them in the suggested bibliography. ↩︎
3. Farrell, T. S. C. (2015). (2015). Promoting Teacher Reflection in Second Language Education: A Framework for TESOL Professionals. Routledge. ↩︎
4. Aspbury-Miyanishi, E. (2022). The affordances beyond what one does: Reconceptualizing teacher agency with Heidegger and Ecological Psychology. Teaching and Teacher Education, 113, 103662. https://doi.org/10.1016/j.tate.2022.103662 ↩︎
5. I am thinking especially of my Language Education for Refugees and Migrants cohorts, who have to write extensive reflective notes after each lesson in their teacher placement. They think it’s hard work writing them (I have to read dozens!). ↩︎
6. Edge, J. (2002). Continuing cooperative development: A discourse framework for individuals as colleagues. University of Michigan Press. ↩︎

---