Episode # 71

Preparing Students for AI in Education with Dr. Laurah Turner

May 6, 2025

About This Episode

On this week’s episode, we are joined by Dr. Laurah Turner, Associate Dean for AI and Educational Informatics at the University of Cincinnati College of Medicine, to discuss the usage of personalized learning and AI coaches to enhance educational experiences.

Guests

Dr. Laurah Turner

Dr. Laurah Turner serves as the Associate Dean for Artificial Intelligence and Educational Informatics and Assistant Professor in the Departments of Medical Education and Biostatistics, Health Informatics and Data Sciences at the University of Cincinnati College of Medicine. She is also the founder and director of 2-Sigma Labs. Her research focuses on applying artificial intelligence and statistical techniques to enhance medical education, refine assessment methods, and address training disparities.

Transcript

Kara  0:00

Today’s episode is sponsored by Learning Blade. Middle school is the ideal time for career exploration. With Learning Blade, you can easily integrate STEM and career awareness concepts into your classroom. Find out if this resource is free in your state at learning blade.com thanks, Learning Blade.

 

Hey. Is this thing on? Are we recording? Can I get a tech person?

 

Caryn  0:29

Oh, for the love of Ed Tech.

 

Kara  0:32

Joining me today is Dr. Laurah Turner, who serves as Associate Dean for artificial intelligence and educational informatics and Assistant Professor in the Departments of medical education and biostatistics, health informatics and data sciences at the University of Cincinnati College of Medicine. So hey, Laurah, how are you?

 

Laurah  0:55

Well. Thank you so much for having me.

 

Kara  0:58

Yeah, we’re I’m excited to talk to you about this actually, because AI has been a huge topic in K 12 education, and so I’m kind of excited to get your perspective being involved in higher ed and your take on different things. But first, can you just tell us a little bit about how you ended up in the educational realm.

 

Laurah  1:25

Yeah, it was, it was a little unintended, exactly how I ended up where I am now. So I got my PhD in biomedical anthropology at Indiana University, and my intention was really to study human evolution and kind of differences in energy and diet and reproductive diseases so

 

Kara  1:57

Interesting.

 

Laurah  1:58

I was doing my dissertation after I graduated was right during the recession, so it was very niche environment, and there weren’t a lot of institutions that were really interested or grant funding locations that were interested in funding very theoretical evolutionary biology, especially In like human evolution at the time. But I did. I did do a lot of concentration in complex statistics and early, very early machine learning, before we even called it artificial intelligence, kind of in the early 2000s when we were kind of getting into more deep learning and the neural networks. And so I didn’t exactly know what I was going to do with it, but I knew I needed to do something. So I started teaching. I got a professorship at NKU in Northern Kentucky University, and then migrated over to the University of Cincinnati and got a leadership position in their teaching and learning center.

 

Kara  2:11

Oh, nice.

 

Laurah  2:46

So I spent a couple of years there really starting to understand, you know, pedagogical theory and building upon that, and had opportunities to integrate a lot of the analytics that I was working on into that. And from there, the College of Medicine offered me an assistant deanship, and so I kind of fell back into kind of the medical round that I was in, and it was a perfect fit for me. And in that role, I would never really did exactly what traditional deans for assessment and evaluations did, because we were always pushing the envelope of trying to integrate natural language processing or big data analytics into how we looked at all of the data. And I was really lucky, because UC is very innovative in that way, and gave me the bandwidth to really explore and the resources that I need to do that. And so kind of keeping with that trend. Once you know, large language models burst onto the scene after the release of chat GBT, my office and my lab was already kind of positioned to just step into that. And so we have been leading a lot of the applied AI in medical education and developing new tools. So then you see actually developed a or created a new position for me, the one that I have now, and an office with the same name and a staff and a lab to continue that work, to really push that envelope and keep us at the forefront, hopefully, of developing and researching these technologies.

 

Kara  4:46

Wow, that’s really exciting. That’s really exciting, but it is kind of like off from your initial thoughts and what you would do with your PHD. So that’s I always love to hear, like everybody’s path, you know, because everybody comes to these little forks, and you go one way, and then before you know it, yeah, you end up

 

Laurah  5:13

in a different direction. Well, I think, yeah, with it coming from a anthropology background was really advantageous, because really, anthropology creates non specialists, and instead, like, how do you take skills or knowledge from one domain, right, or one culture, perhaps, and then apply it in another one? And so we’ve been very successful doing that, taking different techniques that maybe aren’t, you know, traditionally in medical education, and then, you know, translating those into use cases, and then also really creating those, you know, cross disciplinary connections. My lab has, of course, lots of doctors, but also I have PhD students from engineering, from it, from Arts and Sciences and all of that, you know, kind of diversity of thought really helps push and understand how these technologies can be best used in an educational setting.

 

Kara  6:10

You mentioned applied AI, can you explain how you would define applied AI?

 

Laurah  6:17

Yeah, so I think you know, there’s a lot of different AI techniques, but when we’re applying AI, it is, it has to be a specific use case, right? So applied AI and medical education, for example, would be one project that we a tool that we’ve built, which is to create simulated patient experiences using a variety of AI techniques. So we are creating an experience using AI. So we’re applying different techniques to solve a problem, and that problem is, you know, the experience of interacting with patients very early in medical school. There’s not a lot of opportunities to do that, and it’s also very expensive, and so one way to solve that problem is to use artificial intelligence. Potentially, we’re still exploring the utility. So that’s one application. Another application could be something like curriculum mapping, right where we have to align the different assessments with the learning objectives and the program objectives, right? It’s very tedious. It’s very prone to error. People hate doing it, at least the teachers you know and professors I work with. And so it’s a really good use case. And so we’ve created different, what we call AI agents, but you can think of them as bots that go in and do the mapping for you, and then the human just has to review it. So it’s just any time that you’re you’re taking different AI techniques and creating a tool or a system to solve a problem. The problem that you’re solving is the application. And so that’s what a applied AI would be.

 

Kara  7:58

Okay? So how do you see applied AI making an impact in the day to day learning experiences? And how might those innovations inspire K 12 education?

 

Laurah  8:13

Yeah, I think, I think that really, any of the tools that we’re creating, even though they are very niche to medical education can be translated across many different domains, including K 12. In fact, a lot of the ideas, or the foundations of what we build, I test out on my own children. First I have a third grader, and so we’ll, we’ll kind of test certain things and see how she likes it, right? But really, I think the most, the most viable and possibly the most exciting, is the ability to personalize the educational experience in medical education and oftentimes in K 12, it’s very cookie cutter and very rigid, and it’s not flexible for individuals who may have learning delays or may just have a different learning approach or need right and or a different timeline, and that’s really, really challenging, and we know that Bloom back in the 80s, found by looking at K 12, actually, that if you provided every individual with a personalized tutor, you could augment the performance by up to two standard deviations, so ultimately, making a below average student above average or an average student exceptional. But that’s cost prohibitive, and even if we had piles of money laying around, there’s probably not enough humans to actually do that. So could we augment that experience and solve foods two sigma problem with artificial intelligence, we’ve created, or we have created, and we’re we’re testing AI coaches right now. Imagine having your own, you know other, right, or your own knowledgeable other to go through your entire educational experience and help you. To identify when interventions are needed, because a lot of time, I think the gaps in the deficits our medical students feel like they have to hide, and then that leads to, you know, impacts in patient outcomes down the line, because they’re really concerned with being able to match into a residency program. I can see that same thing, where students in the K 12 environment are probably going to be a little anxious about gaps in their knowledge or their skills, but we’ve seen already that students are much more open with the boss than they are humans, because they don’t have that same kind of anxiety around sharing deficits. Now, of course, there’s always kind of that surveillance culture that has to be addressed, and we’re working on really trying to understand where that sweet spot is, but I think that we’ll be able to figure that out and ultimately break down a lot of the barriers that we’ve had before that were either financial or just, you know, human in nature, right? We don’t have enough humans, and we can augment that with artificial intelligence.

 

Kara  11:02

How cool. Okay, so what do you think are some of the key challenges and oper and opportunities when integrating AI? I know you talked about that privacy piece or the surveillance piece, but

 

Speaker 1  11:18

yeah, I think I think that that has been something that really has come up a lot in the field of medicine. There’s a lot of concerns with patient privacy or student data, and so I think that the privacy issue is definitely going to be something that’s very important. It’s not, you know, completely unsolvable or anything like that, but it’s also going to be data ownership. Who owns that data? So all of the tools that we’ve created so far are formative in nature. They are they have no summative component, because that’s part of our our commitment to responsible AI, we know that we’re still in the development of these tools, and so we want to make them safe. The other thing that we’ve done is we have learners co creating all of our tools, and that way they can actually have input into how that data is used and the autonomy of that data. So in our first iteration of a tool that we deployed a prototype, you just kind of entered into the environment and you, you know, did whatever you were going to do since that time, because we’ve gotten feedback, we knew we have found that we need a toggle that allows the learner to say whether or not they allow their data to be used for further improvements or training of the tool. If they unclick that box, then we don’t use any of their data.

 

Laurah  12:57

The other thing for a coach, for example, is we have a toggle where the conversations that the learner is having with that coach can either be fed forward to their human coach, because everything is just to augment the the learning experience, not replace it, or if those are conversations they don’t want fed forward that they just wanted to have, and they want to Keep those private. And so I think that that’s really important. And then also, again, that surveillance, how much is going to be too much? How much autonomy do we need to get if, for example, a learner, we were beta testing this right now, this is something that we came into. The learner was having a conversation with a coach and there was something potentially harmful that was detected. So we do a lot of we do something called Red Teaming, all of our tools, where we actually have learners and our engineers come in and try to break it or do something harmful right to make it, to see if our our guardrails, our safety nets, all of those things work. And so we had a resident talking with the coach, and they started talking about feeling depressed and saying that they were abusing alcohol, kind of as a result of this depression. And we noticed that the way that we had trained it. It wasn’t picking up on it. It was just like, oh, okay, well, that’s interesting, right? And so we had to flag that and go back, and we need to build in guardrails. And then we have to now address the question of, if somebody decides they don’t want to feed a conversation they have with their AI coach forward, however, there is potentially harmful to the user information that is disclosed. How do we handle that? And so we don’t have a good answer for that yet, but it’s something that we have we are actively exploring right now, because this is one of those safety things that has to be addressed that’s going to be very important in. Are, you know, in the medical education environment, but that would be even more important in the K 12 environment, right? And so these types of things are going to be really, really important. Then also, how, how much human oversight Are you going to need? As you know, artificial intelligence is growing at an exponential rate every single week, we have a new model that is even more capable than the in the last so I actually just wrote a paper with Sanjay Desai and a couple of couple other of my colleagues on what we call the alignment paradox. As these AI systems become more and more powerful and more autonomous. How do we ensure that they continue to be aligned with our educational goals and values of our institution? And that’s really important. And so I think those three, those three areas we could go on about, you know, the challenges and opportunities all day, but those key areas, safety, data, privacy and then alignment of educational goals and values are going to be the real areas that we have to tackle. There are other ones, like good use cases, right? So I mentioned that we’re building an AI coach for reflection. It sounds like a great idea, right? We’re not really sure if it’s a good use case. We built it like another human coach. You actually talk to it. You don’t type or anything like that. You actually just have a conversation. But the end user might be like, I don’t want to talk to your AI coach, like, I don’t want to have a mentor. That’s a coach. That’s not something that I like. But we don’t know any of this yet, and so that’s another part of our research is like figuring out what the good use cases are for this type of this type of technology.

 

Kara  16:53

Teachers, you can now earn contact hours toward grad credit for listening to the podcast.

 

Caryn  16:58

Visit for the love of Ed tech.org, for more information,

 

Kara  17:01

What foundational skills like data literacy or ethical reasoning, should K 12 teachers consider emphasizing to their students now as we prepare for this? AI enhanced field?

 

Laurah  17:16

Yeah, so I think that you hit the first two, you know, right on. I think some other ones that are equally important is systems thinking. I don’t think that it is unique to medicine. I think that creating AI, applications and tools, you have to have that systems thinking of seeing how complex components interact within healthcare, within an educational environment, within an ecosystem, if you’re going to go into biology, there’s so many different ways that you can really take a look at why systems thinking would be so important. I think collaborative problem solving is also going to be really, really key. Working with interdisciplinary teams and multi disciplinary teams is going to become even more important. I mentioned kind of the infrastructure of our lab, two sigma labs, and we have, yes, we have a lot of physicians, and we have medical students and residents and fellows, but we also have individuals from it. We have them from engineering, we have them from computer science. We have them all the way from the DAC, the college in the design college for like UI, UX, and every single person brings a unique perspective and a unique way to problem solve that. You know, we all have to work together, because when I can start talking very technically with my data scientists at our lab meeting, but we have to make sure that all of the physicians who are incredibly smart but just don’t have a background in AI understand exactly what we’re talking about, because then they’ll tell us, Oh, that’s not going to work here, and why. Or they’ll ask a question, and it will just make everything so much better. But our ability to communicate right across all the different fields is key in this. And then I would just say the last one, and this is going to be the anthropologist talking in me. It’s adaptability. I think this, this just competency, or skill of being adaptive and developing comfort with continuous learning as technology involves, is really key. I think that our learners, especially in the K 12 environment, do it kind of naturally. Um, but I think that skill is is, you know, needs to be continuously cultivated over the entire educational career of somebody in medicine. We never stop learning, right? It’s called dedication to lifelong learning, and so we’re constantly trying to figure out how we can teach. Teach ourselves the next thing or improve.

 

Kara  20:03

Yeah, I like that one adaptability. Can you share an example of how AI is used in healthcare that would resonate with younger students and help them see the real world impact of what they’re learning in science or technology classes?

 

Laurah  20:22

Yeah? Yeah. So there’s a couple of different ones. I’ll give you an, uh, an older one, and I’ll give you a newer one. So AI in healthcare has been around for for quite a for quite a while, right? Um, so I think that students would be able to relate to the concept of teaching a computer to recognize patterns. And we’ve been doing this in medical imaging like X rays and MRI. This is similar to how they learn to recognize objects right or faces. And yeah, it’s, it’s a simple, it’s a simple thing that could, I could see it involving students trying to identify different structures, not even necessarily in an x ray, but any any image, and then comparing their accuracy to an AI system, a very, you know, traditional kind of machine learning algorithm is always to show pictures of like cats or dogs, right train them and then have them able to identify novel pictures of cats and dogs. But really what this is, is pattern that recognition and it helps understand both the capabilities, but then also the limitations of artificial intelligence. We know that with X rays and MRIs now the algorithms oftentimes will be able to detect subtleties that the humans cannot. And so it’s a really, really good use case. And that’s you know, the imaging specialties like radiology and ophthalmology have integrated these technologies for quite a while. Another example could be a smart insulin delivery system for diabetes patients, again, using machine learning to predict when that insulin is needed based on patterns and blood glucose levels, right, and that would be individualized to the person right, depending on what their diet is and their activity habits, etc. And you know, this would demonstrate how AI can help people manage chronic conditions and improve quality of life. And then I think the most recent one that is really interesting, and we’re starting to really get into, is ambient data capture. So ambient data capture would be, you know, the audio between a physician and a patient, and that’s being integrated now into a lot of healthcare systems at a pretty high and rapid adoption rate. For example, there’s there’s DAX AI, and these act as ambient scribes. So instead of the doctor having to take the notes and or, you know, structure, the entire encounter during or after the AI actually just is turned on, listens to what’s happening in the room and then creates the patient note from the actual data of the conversation that was had. Now we’re we are already trying to integrate this into being able to evaluate or assess things like communication or clinical reasoning using ambient audio capture in our lab. But you can also imagine extending that ambient audio capture into the classroom, where students who are taking notes can focus on different things instead of every single word. Now we have some of this now with, you know, recording of lectures and stuff like that. You can imagine how it could be more personalized to the needs of the actual learner, instead of, again, this kind of like cookie cutter or very rigid system,

 

Kara  23:54

Yeah, from your perspective in higher education, what are some age appropriate ways K 12 educators can begin introducing concepts of AI in the context of human health and medicine.

 

So for elementary students who are really young, I think pattern recognition activities, again, is it’s kind of the foundation of all AI if you think about machine learning concepts and recognizing, you know, patterns of imagery recognition or natural language processing and patterns of speech, all the way to neural networks in large language models, where you have layers that are dedicated To detect, to detecting a certain type of pattern across whatever you’re trying to do. So I think that’s really important and an easy thing that a lot of the students do anyway Middle School. I think wearable health devices right, or wearable ambient capture devices. So our lab right now are working on. AI glasses that capture these. These are things that might be really interesting and understanding how data is collected, and then how that data can be used to make recommendations. You know, my, my daughter, again, she’s started a little bit of running, and so she likes to where, you know my my Garmin, watch when she’s running, and look at, you know, her pace and stuff like that. And we’ve kind of charted across multiple runs and see, is she getting better? Is she getting worse? What is she doing? And so those types of things starting to translate the data and interpret what it means high school students, I think, really could get a little bit more complex and have fun and engage in case studies like and start looking at like ethical dilemmas, right in healthcare, or even not healthcare. A lot of these are, you know, yes, you can put it into that context, but it’s there’s so many other contexts that these could be applied to, right? So, if we’re talking about that ambient audio capture, right? You could think about balancing efficiency with personalized care, right? We could think about like, when is it appropriate to allow a an artificial intelligence system to make a decision about something, and when is it necessary for a human to either be in that loop? And what does that mean? Is the human looking at it and validating it, or is a human’s hands off and there is a completely autonomous system? Or where are the situations where an AI is dangerous and should not be allowed to make any decisions and human is necessary. I think those that’s where, you know, in high school, and then in college, you can start to look at that, and then across all grades, I think just AI tools and literacy and emphasizing that they’re at least currently and in my opinion, for the foreseeable future, AI tools are going to be designed to augment human capabilities, not replace that judgment, which is crucial, and understanding why that is necessary and important, ultimately, will equip our learners to be able to, you know, dive in and use AI tools when they are ready or when they’re presented to them, which I think is going to be earlier and earlier over the next couple of years.

 

All right, well, I would like to say thank you so much for such a great conversation and sharing your insight into all of this in kind of a different world.

 

Laurah  27:36

Thank you so much for having me. This was a wonderful, wonderful experience.

 

Caryn  27:40

Thank you so much for listening to this episode. We hope you found our discussion interesting and insightful. Don’t forget to check out the show notes on our website, where you’ll find more information about today’s guest resources we mentioned, and information on earning contact hours for listening.

 

Kara  27:57

If you enjoyed this episode, please subscribe, share it with your network and check us out at for the love of edtech.org, this podcast is produced by SOITA, the Southwestern Ohio Instructional Technology Association, in partnership with Think TV and CET, the local PBS stations in Dayton and Cincinnati. Until next time, keep exploring new ways to empower your students through tech, and we’ll catch you in the next episode.