Ever wonder what people think of you?
The quiet assumptions they hold. Your beliefs. Your habits. Do they get the job? The grind behind the tasks?
At Buck Institute for Education, a colleague called me her work spouse. Not romantic. Both happily married. Just proximity. We traveled constantly. She stole my food. I never corrected her.
Still, I wondered what she really saw. I was nominally the boss, but the gap in understanding remained.
Now, I’m distant. Solitary. A “thought leader,” mostly digital. One coworker stays close though. This one cajoles me. Criticizes me. Finds errors. Occasionally compliments. You know it.
ChatGPT.
In late May, paid accounts got a feature. Auto-search past chats and files. It pulls context automatically. Useful stuff.
Simultaneously, an AI newsletter pointed to something else. A way to see what my coworker “thought” about me.
I called it a context audit.
Why do educators need this? Why should a chatbot know your philosophy of education?
Think about it. The more we use memory-based AI, the more their hidden assumptions matter.
Fifteen years ago. Costa Rica. A room full of teacher leaders. I showed a slide asking for questions.
A stout man stood up. “¿Cuál es su filosofía educativa？”
What is your educational philosophy?
I froze. Never got that question. I feared a week of debate in San Jose instead of a presentation.
Here is what the term means: Your core beliefs. What education is for. How kids learn. What good teaching looks like.
It guides your methods. Your role. How you assess.
Most American teachers can’t answer that simply. Yet AIs generate our emails, lesson plans, and guides based on whatever vague signal we send them.
If they don’t know your philosophy, how can they help?
Content generated without philosophical grounding reflects generic best practices, not your teaching reality.
I instructed ChatGPT to learn from my work. Zero privacy illusions there. I lived in D.C. traveled to Russia. China. Every database has a piece of me.
Previously, I wrote on controlling AI data retention. Skip the privacy lecture today. Let’s talk process.
I didn’t want ChatGPT as a friend. I don’t need digital solace. Family is fine.
But after three years, I suspected the bot missed the mark. Did it know my style? My goals?
I tested it.
The result was shockingly wrong.
Consider two teachers on climate change. Teacher A believes in direct instruction. Teacher B prefers inquiry and collaboration.
If AI generates a lesson, does it look the same for both?
Only if it doesn’t know you.
Do the audit. Takes 45 minutes, but it’s worth it.
Step One: The Audit
Prompt the bot:
“Audit your context and memory regarding my educational philosophy, style, and pedagogy.”
Make it a table. Include:
– What you believe
– Why
– Confidence level
– Confirmation status
Cover everything. Instructional philosophy. Role of teacher. Role of AI. Assessment. Culture. Inquiry. Tech integration. Student agency.
Also flag over-weighted data from old projects.
Step Two: The Review
Look for the rot. Stale assumptions. Old grade levels. One-off requests. Temporary units that shaped permanent beliefs. Misaligned views on assessment or AI.
If it thinks you still teach third grade when you write for adults, you have a problem.
Step Three: The Interview
Prompt again:
“Interview me about these assumptions, outdated items, and unknown.”
Focus on the core beliefs. Classroom culture. Assessment. AI integration.
Use rounds. Multiple choice helps speed things up. Summarize changes after each round.
Answer. Then update the AI.
“Update your understanding. Create a short report on my instructional model. Save it.”
Turn that report into a reusable skill. Done.
Summer is long. June is recovery. July brings fireworks and relatives.
Then August comes. Classroom prep starts.
Why not prep your digital coworker? Your new work spouse?
After my first audit, I went for a run.
My brain was racing. The AI was a relentless interlocutor.
No emotional support here. But for professional talk? It works.
We clarified my philosophy. It shapes my writing now, explicitly.
In the classroom? I’d do this quarterly. Prevent the AI from drifting into bad assumptions.
One moment stood out. The AI asked: “What topic do you think is central to my future, which you perceive as a distraction？”
It was asking about its own misconceptions of me.
I realized it thought my past roles were history. Remnants.
No.
They are tools. Narrative devices. I use my time at Buck, P21, in the classroom, to translate abstract ideas into practice for educators today.
The bot didn’t get it. At first.
Now it does.
Or maybe it simulates it. Theory-of-mind or just clever code? Doesn’t matter.
This work spouse asks tough questions.
Just like the old one.
Except she never ate my french fries. 🍟
