Meet Dr. Maxwell

Meet Dr. Maxwell: Enhancing Student Engagement through Dynamic AI Simulations

By Tim Whiting

Pat Maxwell. That’s his name. He is a high-level hospital executive to whom you must present your case. You cannot achieve what you need without meeting with Dr. Maxwell. But what do you know about him? You can’t find his bio on the hospital’s website. He has no LinkedIn page. Why? Because Pat Maxwell is the creation of Bailey Freeman, Assistant Professor in the Nurse Anesthesia program at Texas Wesley University. Maxwell is an AI bot created for use in Dr. Freeman’s leadership communication course to simulate real conversations that her students will have one day.  

Bailey Freeman has taken AI and embraced it. She has made it a real part of her classes. Students have had the opportunity to have conversations and experiences that cannot be replicated with a rehearsed role-play situation. Conversations that can only be matched with the actual experience. How did she do it? What lessons has she learned? Where will this take her and her students? The following interview dives into Bailey’s experience and lessons learned in creating Pat Maxwell and his “colleagues” over the last year. 

The following is a conversation with Dr. Freeman, conducted on March 3, 2026. It has been lightly edited and organized for clarity and readability, with the original meaning and substance of the discussion preserved. For a clear understanding of Dr. Freeman’s experience, the essence of the interview remains intact and can be read here: Interview with Bailey Freeman, DNP, CRNA 

The Industry 

How have you seen AI, and, more specifically, AI bots, in the health profession? 

We're starting to see it (AI) a lot more integrated into our electronic health records, when we're doing all of our charting, like in the operating room. Think about clinical decision support as being tools that support how workflow is done. This could be a drug dosing guide or a protocol written on paper. But now those things are integrated into our electronic health records. Now, with AI, we're able to analyze large amounts of data, these workflows, clinical decision support features with that technology behind it. So it is interesting.  

There are also very practical applications emerging. For example, with ultrasound, many of the tasks we use it for—such as cardiac assessment or identifying anatomy for regional anesthesia—are now being supported by AI-guided tools. These technologies are helping make the process more efficient and accessible in clinical practice.

Do you see something similar to what you did already in professional development for healthcare professionals?  

Not yet, but I think there’s a lot of potential—especially with the interview bot, because it’s one of the harder skills to learn when you’re first starting out as a clinician. You have to go talk to people, and if you don’t feel confident or don’t know what to say, it can be really awkward because you’re asking about their personal health history. And it’s not just asking the questions, it’s knowing how to dig a little deeper and actually get answers out of them.  

 Being able to practice in a way that’s not just a one-time assignment makes a big difference. It becomes something you can use anytime—even in the car or just talking through it on your own and practicing that back-and-forth. Because the responses change based on what you say, you’re getting that dynamic interaction, and that’s what really helps build those skills. 

Pat and His Colleagues 

So, tell me how are you using AI bots in your curriculum? 

Yeah, so it all kind of started with the leadership class I taught over the summer. It was my first time teaching it, and I wanted to find a way to make it more engaging. I also wanted students to actually practice leadership communication in a simulation, not just talk about it. At the same time, I was starting to get more interested in AI and was reading a lot about intelligent tutoring systems. I remember thinking, “What if I built an AI bot that students could actually talk to—something that could carry on a conversation and negotiate with them?”  

That’s really where it started. Developing it took a lot of time and research, especially figuring out how to build in guardrails. What I found worked best, based on what I was reading, was a rubric-based approach—essentially creating an internal structure for the AI to guide what it should and shouldn’t do. That was probably the biggest challenge.  

Once I had that foundation, I was able to build two additional bots for other courses with some minor tweaks. 

One was a pre-operative assessment bot where students could practice interviewing a patient. They really loved that one—we used voice mode so they could actually have a back-and-forth conversation.   

The other was for our finance and business management course, where students could pitch a staffing proposal and negotiate in real time. Each of the bots has its own knowledge base, but they all use that same rubric structure to guide the interaction and keep things on track without the AI jumping in to teach too early.  

The Process 

What do you feel was the easy thing about it? What do you feel were the obstacles in there?  

Actually, making it was really fun. The roadblocks came in more during testing—figuring out how to set the rubric appropriately and make sure it was actually working the way I intended. One thing I’ve found with building AI agents is that the AI will kind of help you build it.   

I was going back and forth between Claude and ChatGPT, and that ended up being really helpful because it felt like having two separate conversations that weren’t influencing each other. I could run something through one, look at the output, and then run the same thing through the other and compare what I was getting.  

That was especially useful when I was developing the guardrails and trying to make sure the bot was going to behave the way I wanted it to.  

Did you have a resource that would guide you or did you just kind of have to just figure it out?  

I played with it a lot. I had some other faculty play with it. My husband tried to game him a little bit. I think people were kind of having fun with it, you know, trying to make him break character or say things he wasn't supposed to say. At the same time, I was grounding it in evidence from intelligent tutoring systems and clinical simulation guidelines. 

So, which AI tool did you find overall was the better one to use?  

At the time, I would say ChatGPT worked really well because you could build a custom bot, and it was easy to use and free. That made it accessible for students—I could have them sign up with an email and start using it right away. 

We did run into some limitations, though. After about 10 prompts, it would require a paid subscription, and we didn’t want students to have to pay just to complete an assignment. So we worked around that by putting students into groups where at least one person had a paid account. That was a bit of a roadblock for the simulation. 

As far as building it, I originally used GPT-4, which at the time had a tendency to be a bit of a “yes man.” It wanted to teach and explain everything. So if you asked something like, “Hey Pat, how do I talk to you?” it would just tell you exactly what to do. I had to spend time figuring out how to structure the prompting so it wouldn’t do that. That took some trial and error.    

The Response 

What was like the very the initial response of students?   

I think they thought I was a little crazy at first. When I introduced the assignment, there was definitely some hesitation—like, “Okay… what is this going to be?” They didn’t really know what to expect, and a lot of them were nervous.  

It actually felt very similar to real life for them. That sense of, “I’m talking to an executive, I don’t know what he’s going to say next, and I don’t know how I’m going to respond,” which creates that same kind of anxiety you’d feel in an actual conversation. And that’s really what I was trying to simulate, because there’s not a great way to recreate that. You can use actors, but this created a more dynamic, back-and-forth experience that felt realistic. 

By the end, though, they were excited. They were proud that they had worked through the interaction and reached an agreement with Pat Maxwell. We did a full debrief afterward and talked through their perceptions, and overall the response was very positive. They had fun with it and wanted to do it again. 

Did they (the students) have frustration with using the AI bot?  

Not really, at least not that I saw. If anything, it seemed to increase excitement and engagement, especially with material they might not have been as interested in otherwise.   

Because we’re clinical people, a leadership class you take over the summer isn’t always the most exciting. But this got them more engaged with leadership communication, which is actually a core graduate competency in nurse anesthesia.    

It was relevant, and I think it got them a lot more engaged in the course. That's my hope. At the end of the day, I really just want them to be more plugged in and excited about what we're doing.   

What was it like when you first said to your colleagues, “Hey, I'm gonna do this?” What were their responses? Did you get some criticism?  

I didn't get any criticism, but I just don't think they necessarily understood what I was trying to do. They're like, “You're going to do what now? What is this going to be like?” I think after the simulation took place then they kind of caught on. It took me another several months to write my paper about it, but once that came out and was published, it helped people understand "Oh, that's what you were trying to do.“ So I think it was just kind of this crazy idea that was hard to explain, but everyone was like ”OK, you know, I'm here for you. If you need anything, let me know.”   

 Have any of your colleagues tried to build their own bot?  

Yeah, so, two of my colleagues. Dr. Brian Cornelius and I worked on the pre-op chat tutor because I had built it, and he asked if we could use it in his health assessment class. He built aspects of it and he would come back to me to refine it or tweak it based on what I had previously designed for Pat Maxwell with the internal rubric.    

The other one, designed with Dr. Scott Shaffer, is named Dallas Maverick. He was used in Dr. Shaffer’s Business of Anesthesia course for a finance and contract negotiation scenario. He gave me the information that he wanted Dallas to know, and it kind of went back and forth between us. “Can you add this? Can you do this?” And he tested Dallas a lot to make sure that he was giving the correct information because he had all kinds of numbers and financial stuff. That was a new barrier. But it worked out. I mean, the students have really enjoyed it.     

The Lesson 

I'll be curious with a year later, with different students, and with different technology how much more accepting we are of AI and activities like this.  

Yeah, exactly. The newer class (2028) is a lot more savvy. The class that originally did the simulation—the class of 2027—was already in school during that big rise of AI, so they didn’t have as much time to explore it or get comfortable with it.  

But the incoming students, the class of 2028, have had more exposure to it before starting, so they’ve had more time to play with it and get familiar.  

So yeah, it’s definitely different, it’s a different landscape even just a year later. 

I'll be curious if you'll even be able to predict some of the bot's answers, or as you use this and it progresses, do you think the bot might learn from itself?  

Yeah, I’ve actually already started to see that. I began working on my original bot, Pat Maxwell, a little over a year ago, and AI has improved so much since then that every time I go back to build something new, it’s noticeably better. It almost feels like you have to keep going back in, tightening the guardrails, and tweaking things as the technology evolves.  

I’m teaching the leadership class again this summer, and I don’t want to just repeat the same experience—especially since those students have already had some exposure to the pre-op assessment bot. Last year, I collected pilot data looking at student perceptions and their ability to apply leadership communication techniques during the simulation. We had really positive results, and they wanted more, especially because it involved AI.  

This year, I’m interested in validating the rubric—making sure that what I’m putting into it is actually producing the intended output. I may look more closely at the chat transcripts and analyze them from that perspective. 

I also have a few research questions I’m still working through. Right now, I’m doing a scoping review with my husband, who’s a professor, and my sister, who’s a librarian, to see what guidelines exist for intelligent tutoring systems and AI teaching agents. The idea is that anyone can build a bot and say, “This will teach you X,” but we want to understand how to make that more structured and how to demonstrate that it’s actually doing what it’s supposed to do. We’re finding some really interesting things—it’s just a big process.   

The Next Steps 

Where do you see yourself moving forward with this?  Do you think Pat will get a girlfriend?  

Maybe. You never know.   

Well, I have to get a little further down the road with the scoping review. It's a big process, but I want to complete that. I want to move forward with Pat Maxwell. I would like to continue refining the work that I've done already and continue improving the pre-op assessment bot. I think it deserves more time and attention. I think it could really develop into something really cool for our program, especially being able to talk back and forth with it and practice those skills.  

 
Conclusion 

Initial responses to AI, especially those in higher education, followed one of two paths. Either “Yes. Finally. Life is good” or, “No, no, no, no, never, never, never, no!” Dr. Freeman has found a path to incorporating AI into her course in a manner that stimulates students’ critical thinking while preparing them for a future in health professions. 

Resources 

Access Dr. Freeman’s article: AI-Enhanced Simulation for Leadership Communication in Nurse Anesthesia Education: A Mixed-Methods Pilot Study 

Access the full interview: Interview with Bailey Freeman, DNP, CRNA