What AI can learn from teachers
AI is best able to support professional learning when it is grounded in teachers’ expertise and experience
Author
Partner
Teachers today wear many hats. They design and deliver lessons that meet curriculum expectations while managing classrooms of students with different abilities, circumstances, and needs. They are also expected to prepare children for the 21st century, fostering critical thinking, awareness of diversity, and care for the environment. They design engaging activities and navigate difficult conversations about sensitive topics. And through it all, they try to nurture children’s curiosity and love of learning. With so many demands on teachers, effective professional development plays a crucial role in helping them strengthen their practical skills and meet expectations with confidence and consistency.
Ideally, professional development is relevant, sustained, and responsive to teachers’ needs. But demanding school schedules can limit the time and flexibility available for professional learning, while teachers in under-resourced schools often have limited access to high-quality opportunities to advance their skills. In many cases, moreover, professional development is not sufficiently personalized to meet teachers’ diverse learning needs. AI could open up new possibilities for addressing some of these challenges, making professional learning more accessible and tailoring it to individual teachers.
“AI could open up new possibilities for addressing some of these challenges.”
How AI could support teachers’ professional learning
AI-enabled professional learning can take different forms. It can provide personalized feedback and coaching, and it can help teachers reflect on and respond to classroom challenges. A recent review by the National Institute of Teaching (NIoT) explored the potential benefits of these approaches.
Melissa Bond, one of the researchers behind the review, explains. “It could really help teachers by allowing them to rehearse their decisions and reflect on their practice, enhancing their confidence, self-efficacy, and understanding of professional concepts. Virtual simulations based on real-world scenarios for pre-service teachers, for example, can be helpful in preparing them for the classroom, particularly in navigating tricky behaviours or situations.”
For practicing teachers, AI could make feedback on their classroom teaching more readily available. Video coaching already allows coaches or mentors to observe lessons remotely. At NIoT, a team led by researcher Bronte Mckeown is exploring whether the technology could extend these benefits further with a tool that securely records lessons and analyses transcripts of the recordings, tagging specific teaching moments for mentors to review and discuss with early career teachers.
“But how effective AI is in supporting professional learning depends on how well it can account for the highly contextual nature of teaching.”
“We already know from the literature that video coaching has benefits including improved self-reflection and more specific feedback. And practically, because of timetable constraints, a mentor may only see a given class at the same time every week because that’s when the mentor is free to observe. With the recording tool, teachers can record different classes, giving mentors a broader view of their teaching. A mentor can’t stop and watch every moment of every lesson, but a technology that is able to identify significant moments would provide a mechanism to scale this up,” explains Mckeown.
AI could make professional learning more accessible and responsive, giving teachers new opportunities to practice, reflect, and learn from feedback. But how effective AI is in supporting professional learning depends on how well it can account for the highly contextual nature of teaching. What works in one school, classroom, subject, or country may not work in another, and the needs of individual teachers and students differ. This is where teachers’ expertise and knowledge of their own contexts become essential to making AI more accurate and relevant, and ultimately more useful.
Why AI needs to be grounded in teachers’ expertise and context
“AI could be useful to teachers in generating and refining lesson plans, for example,” Bond says. “But the tool might be trained on data from a context that is completely different from a teacher’s own, so its suggestions might not work in that teacher’s classroom. Working with a professional facilitator or colleague to reflect on these suggestions and consider whether they apply to a specific teacher’s practice and evaluating if they are effective can therefore be really important.”
Keeping people including mentors and colleagues involved when using AI is important because they need to interpret feedback within a teacher’s particular context and identify ethical issues that the technology may overlook.
“Experienced teachers have rich professional knowledge that we don’t want to lose, as well as trust and understanding built through mentoring.”
Bronte Mckeown
The value of combining automated feedback with human, expert guidance was consistently underscored by teachers and schools consulted by the NIoT team developing the automated feedback tool. “We absolutely need to keep mentors at the heart of the design,” Mckeown shares. “Experienced teachers have rich professional knowledge that we don’t want to lose, as well as trust and understanding built through mentoring. AI models can be inaccurate so keeping experienced educators involved in the creation and delivery of feedback is incredibly important.”
While the potential of AI in professional learning is promising, there is only limited evidence of its impact on students and teachers. Understanding what works, for whom, and under what circumstances will require robust research alongside the continued development of these tools.
The potential long-term impact of AI for teachers
Ongoing research will hopefully eventually identify where AI-enabled professional learning falls short, inform improvements, and build a clearer picture of its longer-term impact on teachers and importantly, the students they teach. But this kind of research can be difficult. “It’s expensive, time-consuming and resource-intensive to do at scale. To generate strong evidence, you need a lot of teachers involved and for a long period of time,” Mckeown explains.
The Teacher Education Dataset from the NIoT could help researchers address this challenge. It will bring together anonymized existing data on student learning from schools in the NIoT network and lesson transcripts from the automated feedback tool the team is developing. Over time, this could help researchers, at scale, understand which teaching practices are most closely linked to positive outcomes for students and use the insights gained to improve future professional learning programs.
“We need to understand how we can best utilize each other’s strengths to create truly effective AI-supported professional learning and ensure that robust research is happening.”
Melissa Bond
Building stronger evidence also requires collaboration. “I think a lot more collaboration needs to happen among policymakers, developers, institutions, funders, and educators,” Bond adds. “We need to understand how we can best utilize each other’s strengths to create truly effective AI-supported professional learning and ensure that robust research is happening.”
AI has the potential to make professional learning more accessible, responsive, and tailored to teachers’ needs. But realizing that potential will depend not only on advances in the technology, but also on the teachers, mentors, and researchers who supply the context, expertise, and evidence.
Author
Partner
Footnotes
This article draws on the “Artificial intelligence enabled professional learning” strand of the National Institute of Teaching Professional learning toolkit. This online resource summarizes international research on approaches to teachers’ and school leaders’ professional learning. Combining research findings with insights from experienced educators, it aims to help those responsible for professional development make evidence-informed decisions and improve the quality of professional learning.
Reference for the toolkit strand:
National Institute of Teaching (2026). NIoT Professional learning toolkit: Artificial intelligence enabled professional learning strand.