Will artificial intelligence transform the way teachers are trained? At least one preparation program is banking on it.
The Relay Graduate School of Education, an accredited not-for-profit preparation program with locations in about a dozen states, is developing several AI-driven simulators that will give prospective teachers a chance to practice interacting with students—before they actually set foot in a classroom.
“In order for a teacher to become great, they need high-quality practice,” said Lequite Manning, the department chair of clinical practice and residency for Relay, at the SXSW EDU conference here. “We see AI as an engine for getting more educators more opportunities for high-quality practice and feedback.”
But this isn’t meant to take the place of real, live student-teaching.
“What we envision is an approach to teacher preparation that is both the standard student-teaching in real rooms with real kids and real colleagues, ... all supplemented with very regular opportunities to engage in skill-building in a fully simulated environment,” said Mayme Hostetter, the president of Relay.
At SXSW EDU, Manning and Hostetter demonstrated a prototype for one “virtual classroom” that Relay has developed with Wharton Interactive, a project of the Wharton School of the University of Pennsylvania. This text-based simulator focuses on an early skill teacher-candidates need to learn—getting to know their students.
What an AI teaching simulator looks like
A teacher-candidate is first taken to a landing page with a reminder that the tool is AI-driven and some principles to keep in mind. The candidate is then asked to enter their own demographic information, as well as their experience level in the classroom, which will be used to customize their interactions in the simulator.
The candidate is also asked to share their hobbies: “Part of what we’re teaching them to do as new teachers is to incorporate their students’ interests into how they teach,” Hostetter said, adding that the simulator will model how to do that.
Then, the candidate will take a pretest on what Relay calls the empathy interview, or a getting-to-know-you conversation with students. Afterward, the prospective teacher is again reminded that they will be interacting with AI and is given a list of tips—to provide context and ask direct questions, for example, and to keep in mind the AI’s limitations.
The candidate then watches a video of Hostetter and Lacey Robinson, the president and chief executive officer of UnboundEd, which designs professional learning for educators, sharing best practices for empathy interviews.
Finally, it’s time to interact with the AI. The candidate gets an email from their AI “teacher mentor,” named Sheryl Cameron, who introduces herself and shares demographic information of the school where the simulation takes place. The candidate can ask Cameron for advice on building rapport with students. Cameron answers, based on information input by the teacher educators at Relay.
When the candidate is ready, they begin interacting with the “students.” The students’ personas are not built by AI, Hostetter said. They’re based on real kids she and Robinson have taught over the years.
In the scenario, the teacher-candidate sits down at their students’ lunch table and strikes up a conversation. The candidate picks among several choices for how to start the conversation. The kids respond, and the candidate goes back and forth with them a few times before the exchange comes to a close.
The candidate is then asked to reflect on their experience with Cameron, the AI teacher-mentor. The candidate shares their takeaways, what they learned about the students in that exchange, and how they’d incorporate what the kids shared about their interests and strengths into an upcoming lesson.
Based on those responses, Cameron offers some final feedback about what went well and what they could improve. Finally, the candidate watches a video of Hostetter and Robinson where they either say “great job!” or “try it one more time.”
The future of AI-driven simulators
Hostetter said the simulator will be piloted with candidates this summer, but there’s no firm timeline yet for when it will be incorporated formally into Relay’s curriculum.
Meanwhile, Relay is building additional simulators with Kyron Learning, an AI-based learning startup, that cover other foundational skills, such as giving clear instructions to students and setting up classroom routines.
Classroom simulations in teacher preparation are not new, but generative AI is making them more responsive and—it’s hoped—realistic. Lilach Mollick, the director of pedagogy at Wharton Interactive, said she wants to eventually integrate AI voice and video into teaching simulators.
Hostetter said she hopes these AI-driven simulators will become as much of a staple in professional development for the teaching field as flight simulators are for pilots.
“We see the power of AI being able to really create a much more realistic practice opportunity for teachers who have to navigate so many things going on in their classrooms,” she said. “That can be a huge overwhelm for new teachers in particular. The simulator lowers the stakes—[it] allows people to practice getting it right while the stakes are low, so that they get it right when the stakes are high in front of their real students.”
