The Environmental Trade-Offs of AI in Michigan Education

Written by Michelle Gierman, Michigan Virtual AI Ambassador & Avondale School District AI Strategist.

As both a Michigan Virtual AI Ambassador and the AI Strategist for Avondale School District, I spend much of my time exploring how artificial intelligence can support teaching and learning. Working at both the statewide and district levels, one question continues to guide my thinking: as we expand AI in education, how do we balance instructional benefits with environmental responsibility?

AI tools often feel invisible in practice, but they rely on substantial physical infrastructure. Data centers power the systems behind chatbots, adaptive platforms, and content generation tools, and they require significant energy and cooling. As AI use grows across Michigan schools, so does the demand on that infrastructure.

This matters in Michigan, where renewable energy is expanding but fossil fuels are still part of the mix. Every AI interaction draws from that system. Acknowledging this doesn’t mean stepping away from innovation; it means moving forward with awareness.

A recent classroom example in Avondale illustrates what this balance can look like. A middle school student could understand complex ideas in discussion but struggled with grade-level text independently. Instead of lowering expectations, the teacher used an AI-powered platform designed for schools. Tools like MagicSchool and SchoolAI include guardrails that allow teachers to monitor usage and guide student interactions.

The student used the tool to generate a supported summary and guided questions. The teacher could view prompts, monitor engagement, and ensure the AI scaffolded learning rather than doing the work for student. The result was clear: stronger participation, improved comprehension, and more confident use of academic vocabulary. The AI was used briefly and intentionally to remove a barrier, not as a constant presence.

There is an energy cost to any AI use, but in this case, the instructional return was high.

Now consider a different scenario: a class spends extended time generating high-resolution images to decorate slide decks with only loose connections to the lesson. Students iterate repeatedly, experimenting with styles. While creative, the activity adds limited academic value. Image generation is computationally intensive, and repeated use increases energy demand. Here, the environmental cost is higher, while the educational benefit is modest.

These two examples highlight the trade-off. Both rely on the same infrastructure, but the return on energy use is very different!

As educators, we can ask a few grounding questions:

• Does this use of AI increase access for students who need it most?
• Does it improve clarity, feedback, or instructional precision?
• Are there guardrails that support teacher oversight and academic integrity?
• Is the tool being used for convenience or novelty without clear learning value?

AI does have an environmental footprint, but it can also reduce paper use, streamline workflows, and support differentiated instruction. The goal isn’t avoidance, it’s discernment.

Michigan schools are well-positioned to lead here. We can prioritize platforms built for education, engage vendors on energy practices, and provide professional learning focused on purposeful, not constant, use. We can also help students understand how digital tools connect to real-world energy systems.

I remain optimistic. Across both my statewide and district work, I see educators committed to thoughtful, student-centered practice. When we prioritize high-impact uses of AI and move away from low-value, high-consumption applications, we show that innovation and stewardship can coexist.

Balance, clarity, purpose, and responsibility are what will define effective AI use in education moving forward.