Augment, Don't Automate
A Head‑of‑School’s AI Field Test in Multivariable Calculus
This special edition of The Beacon features a guest reflection from Greville Haslam Head of School, Dr. T.J. Locke. Curious about how artificial intelligence might support teachers, Dr. Locke recently stepped into a Multivariable Calculus classroom—not as an expert, but as an observer and learner. Below, he shares insights from this unique exploration into AI’s potential role in enhancing teaching and learning at Episcopal.
I recently undertook an experiment to explore the practical applications of Artificial Intelligence within a real-world classroom setting. My focus was on finding potential benefits for teacher preparation and instruction. My goal wasn't necessarily to revolutionize teaching, but rather to observe, collaborate with an accomplished teacher, and identify where AI could genuinely add value to his already strong practice.
My Goals for the Experiment
My primary objective was to see if I could find valuable ways AI might assist this expert teacher. As an administrator, I hoped to act as a helpful partner, using AI to perhaps uncover benefits or efficiencies he might not have had the time to explore on his own. Beyond helping this specific teacher, I also wanted to better understand the potential and limitations of AI in education more broadly, especially concerning its possibilities for enabling higher-level thinking and fostering teacher-AI collaboration.
How I Conducted the Experiment
I approached this experiment through several stages:
Selecting the Teacher and Subject:
I chose an accomplished, experienced Multivariable Calculus teacher I already had a good relationship with. This relationship was key, as my position can sometimes be intimidating for classroom visits. His mastery meant I could focus on augmenting good teaching, not fixing deficits.
I specifically picked multivariable calculus because I don't understand it at all. This was intentional; it ensured I wouldn't bring my own biases from my background as an English teacher.
The teacher uses a traditional, teacher-centered approach, relying heavily on the chalkboard. This presented an interesting context for exploring AI integration.
Pre-Observation Collaboration: We started with a meeting lasting over an hour, where we discussed his approach and how I envisioned the experiment.
Classroom Observation: I observed two of his full class periods. I decided against audio recording to avoid making anyone uncomfortable. During the classes, I took notes and used AI tools in real-time, exploring potential ways it could help. I used prompts like:
"How do I check for understanding?"
"Is there a practical example of this?"
"What would be a good homework question to give?"
"What might some common misunderstandings be?"
"Where should I go next?"
I even took pictures of the chalkboard work and asked AI for analysis or suggestions.
Post-Observation Debrief: Afterwards, we had a half-hour debriefing session to reflect on what I'd seen and discuss the AI possibilities I had considered.
Results & How It Went
The experiment yielded nuanced results. The direct, immediate impact on this specific teacher's practice was, in my assessment, "just a little bit." Here's what stood out:
Limited Immediate Uptake: The teacher is confident in his methods, which have proven successful over time. He wasn't particularly drawn to AI suggestions aimed at changing his core pedagogical style (like making lessons more student-centered) or saving preparation time. He seems to genuinely value the process of crafting problems himself, even if it fills his weekend time. His deep content expertise also meant alternative explanations from AI weren't always necessary.
Areas Where AI Showed Potential:
Visual Aids: He agreed that AI-generated visuals could be helpful, acknowledging the difficulty of sketching complex 3D concepts accurately on a chalkboard.
Real-World Examples: I found AI was surprisingly quick and effective at generating practical applications for the highly theoretical math concepts being discussed. He seemed open and to these and even excited.
Efficiency: Although not a priority for him, I could clearly see the potential for AI to save significant time in lesson prep, like creating problem sets.
Checking Work/Logic: I proposed using AI to review teacher notes, check problem-solving logic, identify potential student misunderstandings, or compare the AI's approach to his own.
Impressive Traditional Pedagogy: I was frequently struck by the teacher's skill, especially his methods for checking conceptual understanding ("elite," as I noted) and the effectiveness of his straightforward approach using simple examples that "force the imagination to wrestle with the math.” There were many moments where I felt AI would not add anything, and might even detract from the powerful, albeit "old school," learning happening.
My Reflections
This process led me to several personal and broader reflections:
The Power of Collaboration: Just the act of me, as an administrator, engaging deeply with a teacher about his practice through extended observation and focused conversation felt incredibly valuable. It's something we have too little time for in education.
AI's Potential Reaffirmed: Even though this teacher didn't adopt many suggestions immediately, the experiment reinforced my conviction that AI holds significant potential for education. I see possibilities for pedagogical variety, efficiency, and maybe even enabling higher levels of thinking down the road.
Appreciating the Human Element: The experience was a good reminder of the effectiveness of strong, traditional teaching methods and the irreplaceable value of human interaction and intuition in the classroom.
The Benefit of Being Unbiased: Choosing a subject I knew nothing about was definitely the right call; it prevented me from imposing my own pedagogical preferences.
Planting the Seed: Even though the teacher only saw small benefits of AI for his practice at the time, he has already taken an AI mini-course for higher ed math teachers, and he is using AI to help him plan some lessons that we will be using as a guest teacher in colleagues’ classes.
Lessons Learned & Takeaways for Fellow Educators
Based on this experience, here are some takeaways I'd share with other educators exploring AI:
Context Matters Greatly: How useful AI is depends heavily on the individual teacher's style, needs, subject, and openness to change. What works for one might not for another.
Think Augmentation, Not Just Efficiency: Yes, AI can save time, but its potential as a "thought partner" is perhaps more exciting. It can help us reflect on pedagogy, brainstorm alternative approaches, generate examples, create visuals, and devise new ways to check for understanding.
AI Can Bridge Concepts: It proved useful for connecting abstract mathematical ideas to real-world applications and could likely help generate varied assessment questions.
Value the Collaborative Process Itself: Simply engaging in structured observation and reflection about teaching practice is powerful. Using AI as a tool within that process can enhance it.
My Advice for Trying This: If you want to do something similar, start small with a colleague you trust. Don't expect to solve everything at once. Let the learning guide you, don't force AI where it doesn't fit, and be willing to experiment. Even analyzing a photo of classwork with AI can spark interesting ideas.
The Future is Collaborative: I'm excited about the potential for AI to enhance how teachers work together. Imagine teams using AI as a partner for instructional design or creating richer assessments.
In the end, while AI didn't revolutionize this master teacher's classroom overnight, the experiment was incredibly valuable. It highlighted numerous ways AI could support educators, underscored the power of focused, collaborative reflection on our craft, and reminded me that we are truly just beginning to figure out how best to weave these potent new tools into the fabric of education.