Keeping the Human in the Loop
Last week, I had the opportunity to attend a conference at Stanford focused on AI use in math education. I came in curious about the logics we use to design Edtech tools—and I left even more convinced that we’re at a critical moment for shaping how AI gets integrated into teaching and learning.
We are still calibrating:
How much AI do we want in classrooms?
What types of cognitive labor are worth automating?
And what kinds of human thinking should always be preserved?
What We Explored
At the conference, I saw several AI tools used to analyze student and teacher discourse—helping to make visible the kinds of patterns that might otherwise go unnoticed. Others focused on feedback generation or on surfacing themes in student thinking. But what struck me most were the conversations about the cognitive work of teaching and coaching.
In my view, some of the highest-leverage cognitive labor in education includes:
Understanding student thinking—not just what students know, but how they reason
Designing student-facing materials that guide and deepen learning
Facilitating connections with families that support relational trust
These are things AI can assist with—but should never replace.
What AI Is Good At
It helps to understand that AI-powered language models, like ChatGPT, operate through a quantitative relationship to language. They identify patterns in symbols and strings of text. That makes them excellent for:
Reformatting content
Generating variations of prompts or lesson ideas
Surfacing patterns across student reflections
But not for interpreting student emotions, building relationships, or making final instructional judgments.
AI is a useful tool for sourcing ideas, finding themes, and reframing messages—but the meaning-making must stay human. It should not replace human thinking—it should deepen it. That principle grounds how I approach learning experience design and continues to guide how I integrate AI into both classroom practice and tool development.
Responsible Use in Classrooms
One of the most powerful stories I heard came from a teacher whose 8th graders were using generative AI to help with open-ended tasks. These students weren’t hiding it—they were excited to show their teacher how they were thinking with AI. Because the teacher gave them permission and stayed engaged, she was able to coach their usage and expand their reasoning. It was a beautiful example of co-agency and trust.
A growing concern for me—emerging from themes at the conference—is the use of AI to automate relational dimensions of teaching. In particular, tools that attempt to infer student sentiment or make culturally responsive decisions without direct human engagement. This raises a red flag: predicting emotions or cultural needs without authentic input risks reinforcing bias and undermining trust, rather than fostering equity.
So here’s a question I think all designers and educators should ask:
Is this something I could ask a student or family directly?
If yes, then that’s where the insight should come from—not an AI guess.
Opt-In Only (For Now)
AI tools are already in schools. And their presence will only grow. But given how nascent this technology is—and the environmental and ethical questions it raises—I believe any student-, family-, or teacher-facing interaction with generative AI should remain opt-in. That includes classroom use. Families should always be able to say no, and so should teachers.
We also know that generative AI has a significant environmental footprint. Until we see stronger commitments to sustainable compute—the practice of minimizing the environmental impact of AI and computing infrastructure—we should avoid implementing large-scale, districtwide AI tools for student use.
What’s Next
This conference reminded me of something essential: we are the ones shaping the future of AI in education. If we want these tools to empower learning—not extract value from it—we have to stay alert, stay imaginative, and stay in dialogue with each other.
I left with new ideas, new connections, and a renewed sense of purpose. Let’s keep building.