Field M. Watts, Lei Liu, and Amy Midyett
Recent discussions surrounding AI in education reflect both enthusiasm and growing concern. Educators, families, and policymakers are raising important questions about the overreliance on technology, risks to student privacy, and potential erosion of critical thinking (Forsström et al., 2025). These concerns are increasingly shaping school policy. Recent Education Week reporting highlights a fragmented policy landscape in which some districts continue to restrict AI tools, while others are rapidly developing governance frameworks in response;to concerns about academic integrity, privacy, and instructional quality (Prothero & Vilocarino Jennifer, 2025). At least some states now require district-level AI policies, reflecting the urgency and uncertainty surrounding responsible implementation.
However, removing AI from classrooms entirely may be a disservice, as students are already using AI tools for schoolwork, often outside of school oversight and through general-purpose systems not designed to support learning (Doss et al., 2025). This tension highlights the need for schools to shift from debating whether AI should be permitted to determining how AI can be effectively implemented to foster student learning. As Crippen, Zhai, and Lee (Crippen et al., 2026)argue, the critical issue for educational systems is the need to establish a clear educational purpose for any AI applications implemented in the classroom.
This tension is particularly important for research and development efforts such as the GENIUS Center’s work on AI-supported science learning. If AI is introduced merely to automate scoring, replace instructional judgment, or undermine students’ critical thinking by providing direct answers to homework or activities, skepticism from parents and other community members is understandable. Many of the current concerns about AI in schools stem from these uncertainties about whether AI could replace teachers’ roles or negatively impact student learning.
However when AI is intentionally designed around learning theories and pedagogical goals, it can support scientific inquiry, reflection, and student thinking. In this context, teacher and parent trust is one of the central conditions for responsible adoption. Importantly, trust does not mean unquestioning acceptance of AI outputs. Rather, it reflects teachers’ confidence that AI is being used in ways consistent with sound pedagogy and with appropriate transparency, boundaries, and professional oversight. While it is still important to establish trust by demonstrating the technological capabilities of AI systems, researchers and developers must also consider how teachers teach with and about these AI systems, allowing for trust to be established through classroom practice.
A recent classroom implementation from the Auto-SPIN pilot study (an IES-funded study introduced in a previous commentary)(Liu et al., 2025) provides a concrete example of what this can look like in practice. In one middle school science classroom, a teacher used an AI-supported feedback tool designed to help students strengthen their scientific argumentation. Her implementation directly addressed several of the concerns that often drive anti-AI sentiment. Instead ofpresenting the AI as an authority, she explicitly framed it as a thinking partner whose feedback should be questioned, interpreted, and evaluated. From the outset, she set clear expectations: unlike more general-purpose AI, the AutoSPIN AI system was designed to provoke thinking rather than provide direct answers. Students initially found this frustrating, noting that the AI “pushed them to think.” However, the teacher used students’ frustration as an opportunity to clarify that productive struggle and reflection are essential parts of science learning.
The teacher also addressed concerns about AI reliability and transparency by making the system itself an object of inquiry. For example, the teacher asked students to submit the same response and compare the feedback they received, highlighting the variation in AI outputs. The teacher used this as an opportunity to spark discussion about how AI systems work, why outputs may differ, and why users need to exercise judgment when interpreting AI feedback.
Most importantly, the teacher maintained clear instructional authority throughout the lesson. Students were encouraged to critique the AI output and decide how to use it to improve their work. In this way, they remained active participants in their learning. The teacher even invited comparisons between the AI feedback and her own, reinforcing the idea that AI should complement, rather than replace, critical judgment.
This classroom example illustrates that trust in educational AI is not a feature of the technology itself, but something that is constructed through classroom practice. As debates about AI in education continue, the path forward is unlikely to be defined by blanket adoption or outright rejection. More likely, it will depend on whether principled models of use can be established to preserve human judgment, promote critical thinking, and align technology with meaningful learning goals.
References;
Crippen, K. J., Zhai, X., & Lee, O. (2026). Responsible and Ethical Principles for the
Practice of AI-Supported Science Education. In Advances in Technology-Rich Science Education (Vol. 3, pp. 49–80). Springer, Cham. https://doi.org/10.1007/978-3-032-16871-9_3
Doss, C. J., Bozick, R., Schwartz, H. L., Chu, L., Rainey, L. R., Woo, A., Reich, J., & Dukes, J. (2025). AI Use in Schools Is Quickly Increasing but Guidance Lags Behind: Findings from the RAND Survey Panels. AI Use in Schools Is Quickly Increasing but Guidance Lags Behind: Findings from the RAND Survey Panels. https://doi.org/10.7249/RRA4180-1
Forsström, S., Njå, M., Munthe, E., Álvarez-Galván, J.-L., & Houldsworth, L. (2025). The impact of digital technologies on students’ learning: Results from a literature review. OECD Education Working Papers, OECD Education Working Papers. https://doi.org/10.1787/9997E7B3-EN
Liu, L., Watts, F., & Zhai, X. (2025, October 1). Building Teacher Trust in AI through Co-Design – GENIUS Center. https://ai4genius.org/uncategorized/building-teacher-trust-in-ai-through-co-design/
Prothero, A., & Vilocarino Jennifer. (2025, October 20). How School Districts Are Crafting AI Policy on the Fly. Education Week. https://www.edweek.org/technology/how-school-districts-are-crafting-ai-policy-on-the-fly/2025/10
