“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
The high-quality practice is to involve skills teachers need such as "getting to know their students." This happens through some text-based interactions. The teacher candidate plugs in some personal info, then watches a video of a talk between the head of Relay, Mayme Hostetter, and Layce Robinson, CEO of UnboundEducation, a teacher PD outfit.
Then the AI finally shows up in the form of a letter from a "teacher mentor" who gives the candidate student demographic info. The candidate can ask the AI for advice on dealing with students. Let me say that again--the human prospective teacher can ask the computer for advice about how to deal with the imaginary human children.
Then the teacher begins "interacting" with the imaginary students. Note--the students are NOT AI, but are personas based on "real kids" that Hostetter and Robinson taught. Robinson have had some actual classroom experience (though only a couple of years are listed in her LinkedIn profile); Hostertter spent two years in a private boarding school, and three years in a KIPP charter. The candidate imagines sitting down at the student lunch table (an interesting choice, that) and "strikes up a conversation." Will says that the candidate has several choices about how to start the conversation, suggesting maybe that what we're actually talking about is a multiple choice Talking To Students quiz. Then the candidate shares with the AI mentor, and then is finally rewarded with one of two videos by Hostetter and Robinson--either an attaboy or a try-it-again video.
So, not very impressive, but one more entry in the drive for classroom simulators for teacher prep. I get the appeal-- an actual classroom is often unforgiving, and when you screw up, you may have to live with the fallout of your bad choice for weeks. But the dream of a classroom simulator that's like a, a Hostetter suggests, a flight simulator, is a silly dream. Simulating a physical object interacting with the laws of physics is a hell of a lot easier than simulating human interactions.
Not that folks don't keep trying. Back in 2016 there was a bunch of noise about TeachLivE, a classroom simulation with CGI students direct from Uncanny Valley School District. This looked creepy, but like this newest wrinkle, it turned out to be far less than it pretended. The teacher trainee is interacting with a CGI rendering of students, but those students were actually animated by "an interactor" who "controls the student avatars in the classroom, speaking through a microphone and using head-mounted and handheld controllers programmed to respond to certain movements." It's a blend of "program control and puppetry."
What classroom simulators seem to have in common is the attempt to look as if they are harnessing cool new technology, when they really aren't.
In the late seventies, I had an education course taught by Robert Schall, who had years and years of actual public school teaching experience. We would develop and teach practice lessons, with our classmates as students. Also in the classroom was "Bobby," a compendium of every annoying student behavior ever. Dr. Schall didn't filter himself through a special algorithm or computer program or even put on some kind of costume. He just sat in the back of the room and gave us the experience of dealing with challenging students. After years in real classrooms, I can confirm that Bobby was an excellent simulation of the real thing.
The best way for new teachers to learn about the classroom is direct experience, and the second best way is from experiences teachers who have spent years in the classroom. Trying to interpose shiny tech is both hugely difficult and also never likely to yield the quality of results from the first two sources.