“We do not learn from experience… we learn from reflecting on experience” – Dewey

At Knowles, we design experiences for our early career teachers that provide opportunities for them to reflect in ways that hopefully improve their practice. This requires us to design a) impactful experiences that provide for rich reflection and b) reflection questions that help teachers productively reflect on their practice so that change can occur. One tool we’ve been using in both design and reflection is Artificial Intelligence (AI). In this blog, I’ll describe how AI aids us in our design of activities and how it helps Fellows in their reflective practice.

My journey with AI began when ChatGPT hit the market in early 2023. Initially, I experimented with ChatGPT before discovering Claude, which I found more aligned with my needs after extensive comparison testing. What started as dabbling with minor tasks gradually evolved into viewing AI as a genuine and useful thought partner in my teaching practice. I’ve found great value in directing Claude (my AI of choice, though this works with any AI) to take on different perspectives – for instance, asking Claude to respond as “a high school algebra student” to help me anticipate misconceptions when I’m working with a new task. Through this collaborative approach, I’ve effectively trained the AI to understand my pedagogical style by providing specific scenarios, sharing my teaching philosophy in prompts, and refining its responses through feedback.

Activity Design

I work with teachers in the first two years of the Knowles Teaching Fellowship. During that time, Fellows meet with me in-person in the fall, spring and summer for a total of 6 in-person days – which amounts to about 48 hours of professional learning. As one might imagine, though we’ve planned for that full 48 hours there are things that come up in the moment that require us to change plans on the fly. During a recent Fellow meeting, one activity had gone long and the next section needed to be cut in half. I had about 5 minutes to make these changes. In the past, I would have just haphazardly cut major sections. This time around, I dumped the original activity into Claude, and entered the following prompt “this is the activity I have planned for my beginning teachers. I now only have about half the time for this activity. Create a new activity for me to lead with about 24 teachers”. Moments later, Claude created an abbreviated version of the section that retained the core goals and activities of the original section, but adjusted times and removed parts of the activities that weren’t as vital. This is definitely something I could have done on my own, it just would have taken me much longer and would have pulled me away from facilitating (which wasn’t an option in the moment).

More generally, there are times when I’m facilitating and I realize that we need a hard break to transition from what we are doing to what we’re about to do. I may not have the time to give the teachers a 20-minute break to clear their minds from the current activity. In those moments, I’ve been able to ask AI to create a quick 5-minute energizer/transition that will help my teachers transition seamlessly into the next activity while also giving them a mental break from what they were just engaged in. Again, this is something I am able to do on my own – just not while I’m currently facilitating. Using AI while facilitating has given me the ability to continue to actively engage with the group, read the room, and be responsive in the moment. In the past I may have asked the teachers to work on a task while my colleagues and I huddle and discuss next steps, outline a strategy, quickly design the next task, and then figure out how to facilitate it. While I deeply value this collaborative approach, I don’t always have the luxury of time or space for this. Using AI gives me the discretion to choose when restructuring needs to be an ‘all minds on task’ effort versus when I can maintain my facilitation role uninterrupted. AI doesn’t replace human collaboration – it creates more flexibility in when and how we collaborate.

Teacher Reflection

At Knowles, we have 3 major goals when working with our teachers:

This requires us to design experiences where Fellows are able to engage in collaborative inquiry, experience equitable teaching practices, and develop leadership skills. But, as Dewey reminds us, it’s not just the experience itself that will help teachers develop their practice, but rather the reflection on those experiences that truly moves the needle for learners. We regularly have Fellows reflect on their learning experiences through the use of reflective prompts, a split-vision mindset, walk and talks, etc. Now, we’re also incorporating the use of AI as a reflection tool. As an example, in Year 1 of the Fellowship, we focus on what it means to be a doer of math and science in our classrooms. We ask our Fellows to reflect on the following questions throughout their first year of the Fellowship:

• What does it mean to be a doer of math or science?
• What does “doing” look like?
• What does “doing” not look like?
• How do I know my students are doers of math or science?
• What part of doing might I put into practice?

We give these reflection questions to our Fellows at the very beginning of the Fellowship to get their initial thoughts and to get them thinking about these questions. We then take them through a math or science task where doing is foregrounded and we are modeling equitable teaching practices – we’re giving our Fellows the opportunity to experience what it means to be a doer of their discipline. We then spend lots of time unpacking what it felt like for them, how they behaved, thoughts they had, beliefs about themselves and others, etc. After giving them the experience of doing, we have them reflect again. We repeat this every time we’re with the Fellows during Year 1 of the Fellowship. Each time they revisit their “Evolving definition of ‘doing’ document, we have them reflect in a new color so that they can see how their ideas evolve over time.

Now, with the use of AI, we’ve made a few changes that have had positive impacts on our Fellows. After each reflection iteration, we ask Fellows to then engage in a discussion with AI using the following prompts:

My Ongoing Conversation with AI:

• Take what you’ve written and paste the entire conversation into an AI of your choice [chat.openai.com, https://claude.ai/]
• Ask the AI some questions (treat it like a true assistant):
  • What might I be missing in my reflections?
  • What might be 3 good next steps for me?
  • What’s something I’m not thinking about?
  • What’s a blind spot that you see?
  • How do you see equity (identity/culture) showing up in my responses? (or not showing up in my responses)
  • How might you answer these questions?
  • Does it appear that I’m writing with a specific population in mind? Who am I not thinking about?
• Add your AI conversation below. This will be an ongoing conversation.

This approach allows Fellows to share their initial ideas, including some that are new and some they are perhaps wrestling with. It also sets the stage for Fellows’ thinking to be pushed. Sharing this rough draft thinking about “doing” with other teachers, can feel risky and vulnerable. It’s how we’ve done it in the past, and there can be defensiveness because of the emergent nature of the ideas. But, having this discussion with AI can reduce the risk and vulnerability while still pushing their thinking and providing feedback. We use this as a valuable next step after reflection but prior to group discussion. This feels safer for some teachers and allows them the opportunity to be wrong, uncertain, or narrow in their scope and revise that thinking before sharing publicly.

What we’ve observed since introducing AI as a reflection tool has been promising. Fellows report several benefits to engaging with AI as part of their reflective practice:
First, the conversation with AI helps Fellows identify blind spots in their thinking that they might not have recognized on their own. The AI can point out assumptions they’re making or perspectives they haven’t considered yet, especially around equity and inclusion.

Second, Fellows appreciate having a “low-stakes” thinking partner before sharing their evolving ideas with colleagues. This intermediate step allows them to refine their thoughts, gain confidence, and come to group discussions better prepared to engage meaningfully.

Finding Balance in AI Integration

As we continue to explore the role of AI in both our program design and Fellows’ reflective practice, we’re mindful of maintaining a balance. AI tools like Claude and ChatGPT serve as valuable assistants rather than replacements for human thinking and connection. The most powerful learning still happens when Fellows engage with each other and with experienced mentors.
That said, AI has demonstrably enhanced our work in two key ways:

1. It extends our capacity as teacher educators, allowing us to be more responsive and adaptive in the moment while maintaining our focus on facilitation and relationship building with Fellows.
2. It enriches Fellows’ reflective practice by providing a safe thinking partner that helps them dig deepers into their experiences, challenge their assumptions, and prepare for collaborative inquiry with their peers.

As with any educational tool, what matters most is how we use it. By thoughtfully integrating AI into our design and reflection processes, we’re finding that it can support rather than diminish the human elements that are central to teacher development. The reflection itself remains deeply personal and necessary, but AI offers new avenues for making that reflection more productive.