Listen to this episode on your phone!
Dr. Mfon Akpan is an assistant professor of accounting at Methodist University. He has a passion for emerging technologies and is an expert in virtual reality technology. He researches new technologies and educational methods to offer students a current, effective, and relative teaching experience.
This is not his first rodeo on the Professor Game Podcast! Find his first interview on
Episode 243.
Guest Links and Info
Lets’s do stuff together!
Looking forward to reading or hearing from you,
Rob
Full episode transcription (AI Generated)
Rob:
Hey, engagers, and welcome back to another episode of the Professor Gaiman podcast. Every now and then we do have some repeat guests, and this is one of those times. Today we have once again, Doctor Mphon Ackban. You can listen to his episode on his first interview on episode 243. M F O N. Definitely. That’s the only interview you’re going to find there. His name is quite unique, at least among our guests. You’ll find that interview as well. If you haven’t listened to it, or if you’re looking at this through our show notes, the link is right there. But Amphon, we need to know, even though you were last time, this time are you prepared to engage?
Mfon Akpan:
Yes, I am, Rob. I’m so excited to be here. This is wonderful. Appreciate the opportunity to come back.
Rob:
Excited to have you here once more. And just as a quick reminder, Doctor M. Von Akpan is an assistant professor of accounting at Methodist University, and he’s been doing all sorts of things today. Last time we were talking a little bit more about virtual reality. This time we are going to be diving deeper into one of those mega trends that we’re listening to these days, but especially how that could be potentially applicable to your gamification and gamified experiences. So, Mfon, we’re talking about AI today. So first off, what are your views on AI, and especially what that means in an educational environments?
Mfon Akpan:
That’s a good question. I think AI is a powerful tool, and I think that we underestimate the power or the usefulness of these large language models. And I think we can tend to focus on the bad side. And what I mean by that is the hallucination, which means it lies.
Rob:
Yeah, I was gonna. I was gonna ask you to clarify, like, I know what it means, but not, maybe not everybody knows what hallucination means in the context of AI.
Mfon Akpan:
Yeah, it’s not reliable. So to give an example, if you pull up your pocket, calculate, or you go on your phone, go to the calculator, you type in two plus two, it equals four. You can do that over and over again and with extreme confidence, you’ll keep getting the right answer. It will work. And you don’t have any trepidation about that. With the large language models, you can work with it. You can ask it nine questions, but then you get to that 10th question and it could give you a completely wrong answer. So there’s an issue with the reliability, which we call hallucination. I think that’s what we tend to focus on, but that could be a win. And we were mentioning that before. That could be something that you can use as a game. Is it telling the truth? Is it working? Can you trust this answer? So it gives you an extra layer if you try to turn that negative into a positive. On the flip side, a lot of the work that they’re doing now is to increase the reliability of the large language models.
Rob:
Yeah. And, you know, in any case, even though the definition of hallucination here is completely different from what some people might have been thinking when they heard it the first time you mentioned it, you know, you still, I understand why they are working to get these quote unquote hallucinations out of the way in these language models, because it doesn’t sound good. That, you know, tool that they’re using to find some answers is not working and is not reliable every single time. So it makes sense, but then it’s a difficult thing to do. I mean, there’s a lot of very smart people working on that. But in the context of education, Mfon, what does AI mean for you? What does it look like? How have you seen it apply? I don’t know, like talk to us about what you’ve seen in the context of education.
Mfon Akpan:
Well, I think from the context of education, it’s a very, very powerful tool when you focus on the opposite side. So if you focus on the negative, all of the things it can’t do, right, it’s not able to do, then, yeah, it doesn’t look very attractive. But let’s peel back the onion and focus on the positives and why you can say it is very powerful, probably more powerful than you would imagine. So when I think about the US and when I look at these measures, so if you go and you look at the benchmark measures for anthropics, Claude, or OpenAI’s chat GPT, it’ll tell you on undergraduate level or bachelor’s level assessments, Claude performs gets 86% of the questions right, GPT gets about 86%. At master’s level or graduate level questions, it gets about Claude gets about 50% right and chat GPT gets about 30% right. Why is that interesting? Well, if you look at education attainment numbers in the US, about 40% of adults in the US have a bachelor’s degree or an undergraduate degree. And you look at the graduate level, it’s somewhere around 13, 15%, and then the doctorate, it’s about 3%, maybe 4% thereabout. So when you think about that, you’ve got something that can perform across at this particular level where you don’t have a majority of the people having that particular level of education. So I think that’s powerful, and that’s something to think about as far as how you can use that in the classroom. The other thing to think about too, in the US is that our literacy rate, if you think about in the US, 54% of Americans read at a 6th grade level. So that means a majority of Americans read below a high school level. So now you’ve got a tool that can write and perform at a good pace at a college level, which is way above majority of the population. So I think using that and having access to that tool is very powerful. Even with the negatives, even with the hallucinations, even with it making errors, it can still be a very powerful tool.
Rob:
So just for me to understand, okay, what I’m getting from that is that you could be potentially thinking of it as a smart colleague or a smarter colleague or somebody you can ask references from. Like, I completely understand that it seems to be at least smarter than a large percentage of the us population for the example you gave. But what does that mean? Like, how can that be useful for the educators or the educateees?
Mfon Akpan:
Because you efficiently. So you have access to a broad depth of knowledge, really literally for free, and you can bring that access to your students. All you need is the Internet. So I was watching an interview with Mira Marati, who’s one of the executives at OpenAI. They’re pushing to get the model to work at a PhD level. And I thought that was interesting. Well, you got to think about it. And this is something, the paradigm. I try to look at it as I’m a doctor of accountancy, so I can operate at accounting at a PhD level, right? That’s what I study. But I’m not a physicist, so if I had something that I can ask physics questions at a PhD level, that’s amazing. And it’s better than I am, right? So it knows more than me. Same thing. When you think if we level set it back to an undergraduate level, okay, you’ve got an undergraduate in american history. Great. You can answer questions in that particular domain. Imagine you have this model that can answer questions in physics, in mathematics, finance, accounting in all subjects, Spanish in all subjects, at an undergraduate level. That’s a powerful tool. And I think that when we think about how smart it is, we focus on the higher end. So we’re constantly saying we want it to be smarter than the majority of human beings or smarter than all PhDs, but, okay, let’s level set and think about how smart is the models that we have right now compared to the general population, then how can we use that in the classroom to help students or give them access to information, knowledge, and I’d say computation that they may not have had already?
Rob:
Interesting. So, in a way, what I’m hearing from that is, and again, I’m not in any way an expert in AI or anything, any of these large language models or any of that very, very interesting stuff, but one of the things I’m hearing from that perspective, and please, again, please, please correct me if what I’m saying is out of what you were trying to convey is like, perhaps, you know, you were saying, well, I’m a PhD in this, but I’m not a PhD in that. So in a way, I mean, it’s not rendering us useless in any way because, of course, to get there, you have to build the model and to continue to build knowledge and so on. But does it look like it’s substituting stuff in education from your perspective or going to be substituting people, tools or other things at this point, like, very, very broadly, like, this is just meaningless at this point, having this thing or this person doing this is just not going to be useful anymore?
Mfon Akpan:
No, I don’t think so. I think it’s only going to aid as far as education. It will make, in my opinion, learning a lot quicker, and it will also add to critical thinking skills, and it will boost the importance of critical thinking skills. So right now, I. I’m writing AI activities. So really most of them are games. So they’re group activities for a company called Wiley. They reached out to me. They said, okay, we have an instructor’s manual. In the instructor’s manual, it gives various activities that the instructors can use in the classroom to gamify or to make the accounting content more interactive and interesting and engaging. So they said, listen, we need to have some AI activities for our principals book and for our intermediate accounting book. So I’m working on that. All of these are group activities, group activities where the students work through the problems, and then they go to the AI. So large language model could be copilot, Gemini could be 40. They go to the model. They have a prompt. They’re asking the model the same question that they work through. So they’ve already worked through this question. Now they’ve got the model doing it. Now, the next step as a group. So they’re working together, and they’re looking to find out, okay, what did I miss? Do we have it right? Is the model right? Are we wrong? Did we miss something. So it’s more of a game of critical thinking, because the model will spit out something, so it’ll give them an answer. They’ve got to analyze it, and they’ve got to work through it. And then they may ask more questions to the model about that answer. Then they go back to what they’ve already done, and then, so it can take what they’ve already done to another level. Because let’s say they work through something, the model works through it. They get the same thing, but the model adds maybe one or two extra things. Now they’ve got one or two extra things that they didn’t see that they can add to and they can discuss and think about. So I think it’ll only add a level of engagement, a level of opportunity, and a level of speed and efficiency that we don’t see right now in the classroom without these types of models.
Rob:
Cool. Cool. So, like you were mentioning what your work is now at Wiley. Could you. What? Wiley is the publishing company, right?
Mfon Akpan:
Yes. Yes.
Rob:
So what is that work that you’re doing right now you’re mentioning? Is it creating, are you creating, learning, learn books or something like that for professors to understand? Something like, what does that look like? What are you doing? And perhaps if you can provide one or two examples, it would be great.
Mfon Akpan:
Oh, so that’s a good question. So you have in accounting for each course. So let’s say an introduction or principles of accounting course, they produce a textbook. So Wiley is a textbook publisher, and they produce a textbook for the students to use, but also for the instructors. They produce an instructor’s manual. In that manual, they have activities, games, simulations, cases at the disposal of the instructor to aid the instructor to give the instructor ideas that they can use in the classroom. So I was asked to create the AI activities. They wanted to have specifically AI activities in the instructor’s manual, which means it’s available to the university. The professors that are using that particular textbook, they’ll have access to not only the activities I’m putting together, but the other activities and cases and games as well for that book. So these are specifically activities that are using AI. And I chose to use free large language models, having group activities with the students. So they work together and they go back and forth with themselves, they interact with themselves, and they interact with the large language models in the activities.
Rob:
Okay, so it’s creating activities that then the students will be doing. So essentially, it’s like creating a learning material that then somebody else is going to use. And you’re letting, of course, the instructors know how to use it because you’re the one who has that knowledge at this point, because you’re, again, you’re the one who created those activities. And is there any very quick reference and not, of course, into specifics? And given that it’s accounting, the specifics are extremely important, I think. But is there anything, any example you can tell us about how you’re using the large language models and AI in these activities?
Mfon Akpan:
Yeah, sure. So, and I’ll give you an example of where I tried to turn a lemon in the lemonade. So, one chapter, we’re doing interest calculations, and then we’re doing the present value of future cash flow calculations and the activity. I have the students work in a group. I give them an example. So they’re all working together, doing the calculations, and then they go to the model to have it perform the calculation, but also in the prompt or in the question it’s asking the model, is there anything else you know, that we may have missed? Can you do an analysis? So it’s asking for some extra information, extra context. But when I test the calculations and really all the models that I use, they’re slightly off. And that’s inherent, and it goes back to the hallucination. Many of the models are not really good at math, and particularly when you start getting a little more and more complex in your calculations, it’s not good in math. Now, the calculations were slightly off, and it had to do with rounding. So it had to do with issues with rounding. So in the material, I naturally correct it. So I have the correct. So, in the instructor’s manual, they have the correct answers, but it will be, and I anticipate a situation when the students are going through it, they may get the correct answer manually, but when they plug it in the model, the answer will be incorrect or slightly off, which is a bad thing. So you may say, well, wait a minute, it’s not doing it properly. The analysis is correct. The definitions, everything else, it’s just the math is slightly off. But I thought that’s a good thing, and it’s a good thing because now the students have to double check, so they have to second guess. So they have to say, well, wait a minute. How did this model come up with this number? And how did we get our number? So who’s right? You know? So now it’s, again, you’re looking to figure out, okay, who’s right? Who’s wrong? Are we right or wrong? How did we get this? So it’ll raise other questions, and I think, again, that goes back to that critical thinking portion of things.
Rob:
Awesome, awesome. Are there any tips that you could give people who are thinking about integrating AI into their teaching? Of course, if it’s about accounting, they’ll definitely take that reference and use what you’re doing on the book. Other than that, if they’re thinking about creating their stuff or thinking something like that, are there any, I don’t know, ideas, thoughts that you have, any best practices at this point? I know it’s very new field and so on, but again, anything, anything like this come to your mind?
Mfon Akpan:
Yes. So particularly in my field, accounting, when you think about the workplace, how it’s being integrated into workflows, you hear the term generate and review. And that means you use the large language model to do a task quickly, and then you have to go back and check it. So you have to review it. Now, the idea is, if there’s a task that would take, let’s say 6 hours to do just straight out your own effort without using any type of AI or large language model, if you generate and review, you can get the task done in maybe 2 hours. So I think one thing to get students to think about is how can they, this is a game. How can you or add it to your workflow, to what you’re doing to improve your efficiency? So when you think about AI right now, the really, the big value with it is in that side of productivity. You can ask it a question, you get an immediate answer. You can ask it to do something, it will do it, you will get an immediate answer. Now, going back to when we talked about hallucinations and it lying, it may not be correct. So you have to review, you’ve got to check what it’s doing. However, in the end of it, you’ll save time. So I think that’s one thing that would be important for students to understand and learn. So another activity I have in my, this is an accounting example. In my auditing course, I have a risk assessment assignment. So students perform a risk assessment on publicly traded company. I give them the option to do that on their own without using AI. Now they have also the option to use AI. And I go through the steps of what you can do and I show them. And when I demo what they can do, and when I demo the whole thing in class, if they use Aih. And then I asked the students, how many of you are going to use AI? I’d say 90%. So the last time I taught this course, I had one student did not want to use AI. All of them wanted to use the AI. And the reason being, after viewing my demo, they saw how fast they could get it done. However, they weren’t aware of the review process. So once they finished the assignment, they said on the front end, it was easy, it went through it, completed this whole, there’s, it’s a whole table they have to work through. They plugged in the information, it filled it out, but they said they had to go back and double check it because when they were reviewing it, some of the information was not correct. But after them going through that activity and going through it, they saw the value. It was able to save them time. So when you compare the student that didn’t do it, it took her a couple of weeks to do this assignment, whereas with the other students, they were able to work through it in a couple of days and get it done with the AI and the reviewing. So I think having the students to figure out ways that they can add it into their work, their day to day, I think that’s a valuable skill, having them work together and finding out, hey, let’s see how long this will take you to do it without it. Let’s try to see. Okay, if you added AI and did this with AI, how long will it take you to do it? Did you save any time? Is it making a difference in what you’re doing?
Rob:
Cool. Cool. So a way of sort of more of a comparison perhaps, of how well this would fare in this or that environment seems to be like one of the good practices that you’ve found in your experience up until this point, right?
Mfon Akpan:
Yes. And then having the students to, I guess you call it freestyle and figure out how they can use, you know, because you’ve got a general tool. So it’s a general tool. You can ask it questions about pretty much anything. It can create images. It can do a lot of different things. So I guess game is to figure out, okay, how can I put this into what I’m doing? How can it help me? And I think that is key, and that’s what’s happening in the workplace, which I think is important in the classroom to get students to think about, okay, how can I use this in the classroom? And then how can I apply this in the workplace? Because definitely what we’re seeing now is you’ve got employees, they may use it at work. However, you got many employees, they’re figuring out ways they may use it at home to do their work. So they’re more efficient, they get things done faster. And I think another thing to think about, it is, you know what? I’ll use this analogy. Think about it. If you had to hire a mathematician, or matter of fact, you had to choose between, let’s say you had to choose between four mathematicians, same education, say everything on paper, their resumes are completely the same. But here’s the difference. Mathematician one, he’s going to do all of his calculations with a pencil and paper. So he’s going to come to work with a pad and a pencil and paper and do all of, you know, the calculation on a pencil and paper. Now you’ve got mathematician number two. That mathematician is going to come with a calculator and is going to do all of the calculations with a calculator. You got mathematician number three. They’re using a computer, so they’ve got a spreadsheet. They’re going on the Internet, they’re using their computer to do their work. Now you’ve got mathematician number four, who’s using AI. So when you think about it, who’s going to be more productive, who’s going to get the work done faster? Right? So is it the person using the pencil and paper, is the person with the calculator, or is the person using the AI? So that’s where we’re looking at things. Well, that’s where things are right now, where you have this bump in productivity by adding AI or using these models or finding ways to use them in your workflows or what you’re doing, even if it’s helping you to write emails, draft documents, if you imagine you’re able to get these things done, maybe it saves you an hour every single day. So that hour every single day, you work five, that’s 5 hours a week. So now you’re ahead 5 hours versus someone who’s not using it. So I think those are some of the things to definitely have students to think about in the classroom, working with it, having them play games, go through it, check their answers, working in groups and seeing how it hallucinates and what it does, well, it gets them to think critically of how they can move forward and use the technology.
Rob:
Cool. Thank you very much for all those recommendations, all that experience that you’ve had using these large language models, implementing this into those learning materials, I think it’s very, very useful to think about it not just as something that’s going to do the work for you, but also in terms of, you know, it’s like a, we’re talking about how much it’s smarter than many people, but then at the same time, the way we have to treat it is as the dumb assistant, right? The assistant can get the work started and do a lot of stuff, but then, you know, you also have to check that work. Are you better off with a smarter assistant? For sure. Right? But this dumb assistant is apparently almost for free, or actually for free. So are you better off with or without the dumb assistant? And most would argue at least you’re even if just slightly, you are better off with this dumb assistant than in this case, is this very smart thing that we call AI. So, Mfon, we are about to take off. Is there any quick final words that you want to leave the audience with?
Mfon Akpan:
Sure. Feel free to reach out if you have any questions. I’m always happy to help. Always happy to help.
Rob:
Cool. Thanks again for that. Also, Mfon, it’s been a pleasure. Loved to reconnect, to get into these new things that you’re working on now. However, as you know, and the engagers definitely know, at least for now, and for today, it is time to say that it’s game over. Hey engagers, thank you for listening to the professor game podcast and I hope you really enjoyed this interview. And by the way, do you have any questions that you would like to ask in future interviews to future guests? If you do, please go to professorgame.com question and ask your question or questions. You can do this several times. If and when it is selected, it’ll come up in a future episode and you will receive your answer by one of our featured guest experts. And remember, as I always like to remind you, because this is really important, this helps us reach more amazing engagers like yourself. Remember, before you go on to your next mission, please go ahead and subscribe or follow. Again, this is totally for free using your favorite podcast app and listen to the next episode of professor game. See you there.
End of transcription
Related
Discover more from Professor Game
Subscribe to get the latest posts sent to your email.
