By Adriana Perusin
Every STEM teacher I’ve worked with wants the same thing: more time for the moments that actually change how students think. The lab where a student’s hypothesis falls apart and they have to rethink. The group discussion where one question opens up three more. But AI in classrooms has mostly meant more screens, not more inquiry. That tension is worth addressing head-on.
The planning burden
Running rigorous inquiry-based STEM lessons means anticipating where students will get stuck, building scaffolding for different readiness levels, designing rubrics that assess process not just product, and sequencing activities so the cognitive demand builds at the right pace. That preparation takes hours. In my 20 years working with teachers across Canada, Brazil, Portugal, and the US, I’ve watched strong educators default to direct instruction not because they prefer it, but because the planning load for real inquiry is unsustainable alongside everything else on their plate.
A different role for AI
The shift I’ve seen working is simple: use AI as a planning partner, not a classroom tool. Instead of putting AI in front of students, keep it behind the scenes. AI can generate differentiated scaffolding sequences, draft formative assessment questions aligned to learning objectives, and build rubric variations for different project types. The teacher reviews, adjusts, and walks into the classroom ready for the range of learners in the room. The teacher arrives prepared to do what only a teacher can do: facilitate thinking.
Key Takeaways
AI works best behind the scenes, not in front of students. The highest-impact use of AI in STEM isn’t student-facing chatbots or automated tutoring. It’s giving teachers back preparation hours so they can invest that time in instructional design and responsive teaching. When teachers use AI to draft scaffolding and then refine it themselves, the quality of what students receive goes up and screen time stays flat.
Good STEM inquiry depends on teacher responsiveness, not scripts. No AI-generated lesson plan survives first contact with 30 students. The real skill in inquiry-based teaching is reading the room: knowing when to let a productive struggle continue, when to redirect, when to follow a student’s question somewhere unexpected. AI can’t do that. But it can free up the mental bandwidth that lets a teacher do it well.
AI can differentiate scaffolding without adding screen time. One of the biggest barriers to differentiation in STEM is time. Teachers know their students need different entry points, but creating three versions of a lab guide on top of everything else is unrealistic. AI can generate those variations in minutes. The student still works with physical materials, lab equipment, and peers. The differentiation happens in the planning, not on a device.
Start small: one AI-assisted lesson plan per week. Teachers don’t need to overhaul their practice. Start with one lesson per week where AI handles the first draft of scaffolding materials. Review it critically, adjust it for your students, and teach it. After a month, you’ll know which parts of AI output you consistently change and which save you real time.
The measure of success is student thinking, not tool adoption. When I train teachers in active learning methods, I always come back to the same question: are students doing the cognitive work? If AI helps a teacher design a better inquiry sequence and that leads to deeper student reasoning, that’s a win. If AI just makes it faster to produce worksheets that students complete passively, nothing changed. The goal is what happens in the room.
Looking Ahead
After training over 1,000 teachers across four countries, the pattern is consistent: teachers don’t resist technology. They resist anything that pulls them further from the work they entered the profession to do. AI that stays in the background and supports the teacher’s judgment rather than replacing it is AI that STEM classrooms actually need.
Adriana Perusin is a Canadian-Brazilian educator with over 20 years of experience in education and over 15 years training more than 1,000 teachers in active learning and social-emotional skills. She founded IASEA in Brazil for teacher professional development and is co-founder of Flip Education.
