Teaching Channel Talks Episode 102: Revolutionizing Teacher Training with AI and Classroom Simulations (w/Mark Atkinson)

Dr. Wendy Amato: Welcome to Teaching Channel Talks. I’m your host, Wendy Amato. And as often as I can, I jump into conversations about topics that matter in education and what’s more important than the intersection between artificial intelligence and ethics. In this episode, I welcome Mark Atkinson, the CEO of Mursion.

Mark, welcome to the podcast.

Mark Atkinson: Well, it’s a pleasure to be here. Thank you for having me, Wendy.

Dr. Wendy Amato: I’m so happy to have you here. I would like for people to understand where you’re coming from and this topic of artificial intelligence. So could you tell me about Mursion?

Mark Atkinson: I certainly can. Mursion is a simulation platform for professionals to practice the kind of skills associated with the work they do under stressful circumstances.

So we work in education, health care and leadership development in the enterprise in education. If you can imagine the use case for imMursion, it would be a teacher. Typically earlier in their career, who is wanting to practice a pedagogical routine. And in a sense navigate the challenges that they would have presenting that content in any subject matter to students who might have cognitive issues around the content, behavioral issues, or social emotional issues, or even language issues in the classroom.

And These might be issues that the teacher’s never navigated before. And in a sense, the practice simulation is an opportunity, as I like to say, to get the jiggles out. Like, let’s go, let’s figure out what we’re going to run into, overcome it, do it under stressful circumstances so that when I actually get in front of real kids, I do this really comfortably.

Dr. Wendy Amato: Mark, I love hearing that because I know I’m not the only teacher who has experimented with that first period class to see how to do it better later in the day. But what you’re talking about sounds a little bit more high stakes.

Mark Atkinson: I think that’s right. I mean, if I put it to you this way, as we think about the ways in which teachers get to process complex pedagogical routines, they’re written down, they might get to observe another teacher doing them.

And that’s an ideal method, or see a video of an expert doing it. it. That’s certainly helpful. But as we know from so many other things we do in our lives, simply watching an expert in sports or in the profession do something to a high degree, it gives us a mental model of what it would look like if we were to do it well.

It doesn’t exactly enable us to do it. Well, there is at the very highest level of learning, the idea of enactment of the new knowledge. That requires adjustment and practice and feedback, I’d say a big part of the simulation experience for us and for our learners is what we call sort of contextual coaching an expert who’s observing you either dynamically or at the recording of what you’ve done to give you sort of the app, what we call the after action review and feedback to say, Hey, Mark, when you asked you know Stephen, the student in the front row, that question, you didn’t notice that Janine on the other side of the room actually had her hand up and had not had a chance to speak for a while in the class.

Those same kinds of things that we know make coaches so powerful in the classroom can be applied towards the simulated experience. And in the simulated experience, you can do no harm to children. So it’s a, it’s an opportunity for you to take risks in a psychologically safe manner. And we think giving people a psychologically safe way to experiment, fail, retake, repractice is a way to get rapid improvement.

Dr. Wendy Amato: I really appreciate thinking about simulations and This word ethics, because some people may think that artificial intelligence or work that we do in the digital world is about compromise or risk. But I hear you saying that working with a simulation is perhaps a strong case for better ethics because we’re not putting real students at risk or someone who’s figuring things out.

Mark Atkinson: Totally. And I would say there are. limitations around a simulated experience that one has to be, as a provider, transparent about, but there are real affordances that, my judgment, overcome those limitations. To be clear, these are digital characters you’re interacting with. It’s pretty obvious they’re not real kids.

However, it is possible to inhabit those Digital characters with either humans or generative AI in such a way that they replicate the kinds of questions that children have about content when taught that content. So imagine a classroom of maybe six or seven children all looking at one another.

Bright eyed and bushy tailed, ready for the lesson, showing up, and able to speak with you in their own personality and voice. But behind all of those children is a human being, somewhere in the world, logging into our software. And through a little bit of trickery and artificial intelligence, we can have those children both talking to one another and to the teacher about the subject at hand with one human being.

So there is already some artificial intelligence in our system that allows us to do that. So that’s how we do it today. and what’s good about that model is that the human is able to recognize in real time the stress that the learner might be going through, the teacher candidate, if you will, going through that simulation.

And ensure that the experience is done in a psychologically safe manner. And this is a huge, hugely important concept around, I think, all of these technologies. They become, I think, powerful tools when deployed in a psychologically safe way. That is to say that you don’t want to. a learner to come into a simulated experience and feel shame or embarrassment.

So you, Wendy, might come in, a truly expert teacher, and we think of the sort of Vygotsky moment as, how do we present the right level of challenge to you to stretch your already existing capabilities and put you on the spot with a little bit of pressure so that you feel that stink. God, I actually hadn’t encountered that problem before.

And so that’s really powerful. If you go through that, you feel, wow, you get feedback on that. You want to come back in and try it again and prove you can do it. Now I might come into the simulation. I’m not you. I’m a beginning teacher. I’m just getting the kinks out of my routines. I’m nervous. So we’ve got, we rely on that human.

To adjust the degree of difficulty of the simulation to ensure that you’re, I’m being challenged, but I’m not being pushed to the point where I feel embarrassed or ashamed. And that’s a trick. I mean, and an important trick is a sort of a trivializing of its importance. We do a lot of training for folks and it’s interesting, the profile of the humans behind our simulations is years of improv training.

And if you know anything about improv training, it is all about being able to put the other person. on the spot without causing them to feel shame and embarrassment. And at the end of the day, we want the learner to have an empathetic experience that’s powerful for them. To get into the theme of your conversation today, I’d say these challenges are easy to navigate with a human.

The challenge in the future is how do you get a gen AI to behave the same way?

Dr. Wendy Amato: Where are we going with this? What’s ahead? Right now you have the expert observers behind the scenes and providing the after action review. Can we get AI to do this work?

Mark Atkinson: I think the answer is not if, but when. And I think the issues around the when are how we balance the same psychological safety with the with a tolerance for error because the machines are still going to make some errors.

Do the humans, by the way, at times the, we think of the cognitive load that our simulation experience presents to the humans behind the simulation. It’s a lot to keep track of. You’re dealing with complex subject matter. You’ve got to remember the profiles of each of the students. You’ve got to adjust for the learner.

You’re scoring the simulation. That’s a lot to keep straight in your head. So I’d be lying if I said humans don’t also occasionally make a mistake. We have a rigorous QA process to minimize that, but nonetheless it does happen on occasion and machines make mistakes. And the question is, are the mistakes that the machine makes tolerable around the valuable values that I articulated earlier of this kind of this zone of proximal development and the psychological safety.

And we’re in the process of learning the answer to that. We don’t know the answer to that just yet. We are, we have been able to build simulations with a single character. And in those cases, it’s quite impressive what we can do. So for example, we could simulate something that teachers never get to practice, which is the real difficult conversation they might have with a parent who comes in upset that their children isn’t performing well at school.

That is one of these big areas of what a teacher has to do on the job that they don’t really get a lot of training and practice on in teacher preparation programs. So that’s something we think in the short term, we probably can get that right in in the short term. In a gen AI type context, much more complicated to teach a state of the art middle school math curriculum in such a way where each child manifests a unique sort of student error or misconception about middle school math.

That’s a little bit more challenging, we’re working on it, but it’s a little more challenging.

Dr. Wendy Amato: My brain is going in all kinds of directions right now. There’s a lot to measure. Could this become something where, you know, your watch is sending data so that the AI knows if my heart rate showing stress because of the interaction.

 Mark, I’d love for us to talk a little bit about privacy. What kinds of things should we be concerned about if we participate in a simulation experience? Are we exposing ourselves to having that data go somewhere

Mark Atkinson: in our model?

We would never share data about you with anyone but you. And I think, that’s really the ethical boundary. We have to maintain. in the same way when we do certain kinds of assessment, we will share aggregated data in The world today, I’m not even talking about a gen AI world with an organization, a school system, a school building to have a sense of how well classes are performing on a curriculum, but not necessarily exposing the strengths and weaknesses of individual children.

There are assessments. that are meant to be identified in the aggregate, and that’s how we feel about these data. Like, it would be very good for a district to know that its middle school math teachers are collectively struggling with these concepts, but there is, there absolutely has to be a firewall around sensitive data about individuals.

And if you don’t create that firewall, no one’s going to take the risk of doing the simulation in the first place, because if the whole purpose of this is to identify my vulnerabilities, I’m not going to, I don’t feel psychologically safe anymore. And I’m not going to take the risks that are necessary for me to actually have the learning we learned that this was a sort of no brainer from the very beginning of our business.

And we’ve adhered to it very strictly since then. We will not share individual data with. Anyone except the learner,

Dr. Wendy Amato: that’s reassuring to hear. I’m sure that people appreciate it. And it definitely changes that first conversation where someone acknowledges that they would like to work on a skill or develop a strength, but not to their own detriment.

Mark Atkinson: Yes, that’s right. Because you want them to be anxious to do more. I mean, the whole point of creating a powerful practice experience is to have people excited to come take more risks, to fail again, but to learn from failure. So that’s absolutely correct.

Dr. Wendy Amato: Mark, help me think about specific skills or knowledge that a teacher can develop using simulation.

Are we talking about delicate subjects, the controversial ones in classrooms today? Or are there other topics we should think about?

Mark Atkinson: I mean, there are some very basic things and then there are some specific things. I’ll tell you the kinds of simulations that we have done. We’ve certainly done things around sort of some basic classroom management type things.

Not the most extreme behavioral things, but just the basic ones about kids turn taking and conversation among students. Navigating interruptions, keeping students in their questioning on task so they’re building on one another’s comments and not sort of diverting the conversation unnecessarily. We do a bit of that.

We do a lot around eliciting student thinking for teachers. How do you, in mathematics and English language arts get a teacher with the right kind of questioning techniques get a student to be very clear about how they thought about the text, what made them think what they thought about the text. In mathematics, the same thing around problem solving.

When they came up with an answer to a problem without necessarily correcting them, how do you really lay bare what the student thinking was? That’s a skill set that we do a lot of practicing and we share out the work. So the student is holding up the example that they’re working on. And explaining how they got to the answer.

And then of course, we do sort of complex pedagogical routines where you do have to navigate what you’re presenting to the students, the questioning and the dynamic conversation all at once, these kind of mini units, if you will. And then we, but we always anchor those very clearly in a specific curriculum.

So they’re about something concrete that the teacher is working on, if that makes sense.

Dr. Wendy Amato: There are a lot of layers in any live classroom, and to think about having the opportunity to focus on one or the other, to prioritize behavior management, to strengthen your discussion guidance, or to think about question asking, that’s really important in developing good teaching.

Mark Atkinson: Yes, absolutely. And we can, the great thing about simulation is we can unpack those things. So, for example, we’ve done a few studies funded by the Bill and Melinda Gates Foundation over the years where we wanted to scaffold teachers towards the effective use of open ended questions. And one of the things that we wanted to see was Could we simply, by working with teachers in simulation, get them to embed more open ended questions in their lesson delivery in such a way, and then start to apply them towards the key concepts in the curriculum?

And there’s no question that when you sort of Isolate an individual skill. It’s very easy to go into simulation and just work on it. And we’ve done other studies since then, funded by other third parties to see, once teachers start to manifest these behaviors in the simulation, do they translate them into the actual classroom?

And the answer to that question turns out to be yes. What that teaches us is that simulation is an opportunity to isolate a specific skill, practice it, and then deploy it and build, and then build skills on top of it. Because to your point. In that classroom example I was giving, there’s language issues going on, there’s behavioral issues going on, and there’s cognitive issues going on, and social emotional issues all at the same time.

And I don’t want to represent that you nail all those things at once. It doesn’t work that way. You try to isolate a variable and work on it and then put it all together. It is still, I mean, it’s the most challenging work there is, let’s be honest. There’s no more work more challenging than teaching.

Dr. Wendy Amato: Educators have been hearing a lot about EQ and having emotional intelligence. Is that something that we can develop through simulation?

Mark Atkinson: I think so. I think so because I think you can unpack some of the behaviors that manifest emotional intelligence. We certainly know it in the adult world, mansplaining is a behavior that can be corrected for.

There’s a Mursion of it that happens in teaching as well, in a way, this this, everything, if you think about the skills associated with wait time, prompting other children to kind of join in the conversation, this way in which the questioning, brings a patience and constructive meaning making amongst the group, though, those types of things, give everyone a chance to demonstrate proficiency and become a sort of the foundational to an inclusive kind of pedagogy.

And I think If I think all of these efforts to bring inclusion to the classroom so that all voices are heard critically important and simulation is a very good way, not just to practice, but to get data on who you called on, how much talking did each digital avatar actually do? Where were your eyes looking?

Where were you paying attention? All kinds of things that tell you things about yourself that you wouldn’t otherwise know.

Dr. Wendy Amato: At the very least, the things that you’re describing are things that I encourage through the use of video in the classroom. But to get the data and the feedback more rapidly using a platform or program, we can really accelerate teacher development and teacher learning.

You’re talking about the things that matter and the things that impact student outcomes. I’m so grateful to hear about it. Tell me more. What would you like for educators to understand about Mursion and about teaching? Simulation and AI.

Mark Atkinson: I, so I would like teachers to understand that there are experiences in simulation that can produce powerful learning in short doses of time.

I, if I were to say I’m a big believer in video and I think video belongs in the constellation of tools that teachers use to learn unquestionably, but one of the challenges around video is the amount of time it takes to analyze and draw meaning out of it. And we know that time is an incredibly precious commodity for teachers.

You can, in a simulation, fail fast and learn. You get to a point you make a mistake, you can immediately go do a do over and get another repetition. And in 30 minutes, it’s like one of those power workouts. You come away feeling, Oh, wow. I hadn’t thought of that. That’s the way I come across. But the experience of the simulation coupled with some of the coaching and feedback at the end is super powerful.

Go apply it to your class. Film yourself in your class, by the way, look at yourself in video. I think they compliment one another very well. Then come back and practice again. And that. Sort of routine that we do in fitness and all kinds of other things in our lives. I think very much works for educators and we’re proud to support them in all of the ways that we can.

Dr. Wendy Amato: Mark, thank you for sharing a conversation with me today. I appreciate getting to have some expertise on artificial intelligence and ethics on simulation and on teacher development. Adult learning is an amazing thing and you are right at the tip of the spear on making some great things happen in education and other fields.

Thank you.

Mark Atkinson: Thank you, Wendy, for your excellent questions and for giving me the chance to speak with you today.

Dr. Wendy Amato: To fellow educators, thank you for joining us in this conversation. If you’d like to explore topics that Mark and I discussed today, please check out the show notes at teachingchannel. com slash podcast, and be sure to subscribe on whatever listening app you use.

It will help others to find us. I’ll see you again soon for another episode. Thanks for listening.