Here’s a question that keeps coming up in my work: why do so many teachers hesitate to use AI tools, even when the tools are available, the training has been offered, and the institutional support exists?
The easy answer is “they don’t know how.” But the research tells a more interesting story. Choi, Jang, and Kim (2023) studied what actually predicts whether teachers will accept educational AI tools. They extended the Technology Acceptance Model (TAM), a framework for studying technology adoption, by adding two variables that matter deeply in education: pedagogical beliefs and perceived trust.
And the findings reframe the whole conversation. The question isn’t just whether teachers find AI useful. It’s whether they trust it, whether it feels manageable, and whether it fits their beliefs about how learning works.
Why Do Teachers Resist AI Tools? The Trust Factor
Trust emerged as a central predictor in Choi et al.’s model. Teachers don’t just ask “does this tool work?” They ask whether it’s reliable, transparent, and compatible with their professional responsibility. AI systems produce probabilistic outputs, operate through opaque algorithms, and make recommendations that affect students directly. There’s inherent risk in that.
Choi et al. found that “teachers’ PT [perceived trust] in EAIT is a prominent indicator for predicting their intentions to use EAITs” (2023, p. 917). Those who trusted the AI systems they were working with were significantly more likely to find them useful and to express intention to keep using them.
I wrote about trust in a previous post on AI literacy for teachers, where Bilbao-Eraña and Arroyo-Sagasta (2025) showed that an 8-hour AI literacy intervention improved awareness and attitudes but failed to build trust. Deeper, sustained engagement with ethics and governance was what it took. Choi et al.’s data confirms why: trust isn’t a byproduct of familiarity. It’s a separate psychological construct that needs deliberate attention.
The study also connects trust to explainable AI. When teachers can see the logic behind a system’s recommendations, their confidence grows. Transparency supports accountability, and accountability is something teachers take seriously. No one wants to hand over pedagogical decisions to a system they can’t interrogate.
One of the strongest findings involves constructivist pedagogical beliefs. Teachers who view learning as active knowledge construction, student-centered, collaborative, and exploratory, were significantly more likely to accept AI tools. Constructivist beliefs positively predicted perceived ease of use, perceived usefulness, and perceived trust.
Choi et al. conclude that teachers are “more likely to accept EAIT when they are more constructivist-oriented” (p. 916). And the logic makes sense. If you already value adaptive, responsive teaching, you’re more likely to see AI analytics and personalized recommendations as aligned with what you’re trying to do. Tools that offer student performance data, identify learning gaps, or suggest differentiated pathways fit naturally within that approach.
Celik (2023) arrived at a complementary conclusion in his Intelligent-TPACK study. Pedagogical knowledge, the ability to interpret AI outputs and connect them to instructional goals, was the strongest predictor of effective AI integration. Technical knowledge alone didn’t get teachers there. Choi et al. add the belief dimension: it’s the pedagogical worldview that primes teachers to even consider AI as a meaningful teaching partner.
Transmissive beliefs, oriented toward direct instruction and content delivery, showed weaker and less consistent effects. Choi et al. attribute this partly to their South Korean sample, which leaned constructivist, and partly to the national AI policy context, which had already shaped generally positive attitudes toward AI in education (p. 917). A good reminder that beliefs don’t operate in a vacuum. Policy climates and institutional culture shape perception too.
Ease of Use Matters More Than You’d Think
Here’s a finding that surprised me. In most TAM studies, perceived usefulness is the dominant predictor of technology adoption. Not here. Perceived ease of use (PEU) came out on top.
Choi et al. report that “teachers’ intention to use EAITs was most affected by PEU” (p. 918). They tie this to limited experience with emerging AI systems. When something is new and unfamiliar, how easy it feels to use becomes the deciding factor. Everything else is secondary.
Park et al. (2023) found something similar in their Singapore study. Science teachers spent most of their preparation time building personal understanding of AI concepts, with little energy left for thinking about how to actually teach with AI. Confidence was the biggest barrier. And confidence is closely tied to how manageable the tool feels in your hands. If a system feels overwhelming or opaque, no amount of theoretical usefulness will get a teacher past that first hurdle.
The design implication is clear. Even brilliant AI analytics won’t gain traction if the interface is confusing, the cognitive load is high, or the onboarding experience is frustrating. Usability comes first.
What This Means for AI Adoption in Schools
Choi et al.’s model explained 69.4% of the variance in teachers’ intention to use AI tools (p. 915). A strong result. And it tells us something structurally important.
Three variables matter most: ease of use, trust, and constructivist beliefs. They interact in a specific sequence too. When a tool feels accessible, trust builds. When trust is there, the tool feels more useful. And when beliefs are aligned, the whole chain gets stronger.
Professional development programs need to go beyond tool demonstrations. Building trust through transparency, supporting constructivist pedagogical thinking, and prioritizing hands-on experience with well-designed systems are all part of the equation. Choi et al. put it simply: EAITs should be “easy to operate, useful, and configured to be reliable” (p. 918).
Usability, utility, and reliability. Worth keeping in mind for anyone involved in choosing or implementing AI tools in schools.
Reference
Bilbao-Eraña, A., & Arroyo-Sagasta, A. (2025). Fostering AI literacy in pre-service teachers: Impact of a training intervention on awareness, attitude and trust in AI. Frontiers in Education, 10, 1668078.https://doi.org/10.3389/feduc.2025.1668078
Celik, I. (2023). Towards Intelligent-TPACK: An empirical study on teachers’ professional knowledge to ethically integrate artificial intelligence (AI)-based tools into education. Computers in Human Behavior, 138, 107468. https://doi.org/10.1016/j.chb.2022.107468
Choi, S., Jang, Y., & Kim, H. (2023). Influence of pedagogical beliefs and perceived trust on teachers’ acceptance of educational artificial intelligence tools. International Journal of Human–Computer Interaction, 39(4), 910–922. https://doi.org/10.1080/10447318.2022.2049145
Park, J., Teo, T. W., Teo, A., Chang, J., Huang, J. S., & Koo, S. (2023). Integrating artificial intelligence into science lessons: Teachers’ experiences and views. International Journal of STEM Education, 10, 61. https://doi.org/10.1186/s40594-023-00454-3