AI isn't new. It's been around since the 50s, going through waves of excitement and funding, then lulls. But now it feels different.
That's why I started this podcast. I talk to people actually building AI - founders, engineers, scientists. Those doing the work.
The goal is simple: show what's really happening in AI.
In this episode, I speak with Dr. Mfon Akpan. He teaches accounting and has worked at several universities. He also helps oversee finances for Fayetteville.
We cover
Dr. Akpan knows a lot about mixing new tech with old-school fields. He's been teaching and researching for over 10 years.
If you want to understand AI's real impact on business and society, beyond the hype, this conversation is worth your time.
Welcome to the pivotal clarity and a I podcast today. I'm thrilled to welcome Dr M for Akan, an esteemed assistant professor of accounting at Methodist University. He's an accomplished expert in financial management with over a decade of experience in teaching, research and service. Doctor Akan brings a wealth of knowledge in integrating new technologies into traditional fields.
Uh, his work extends beyond the academy, and his role as an audit committee member of City of Farewell is his most notable. And he provides critical oversight in financial affairs. Join us as we dive into how a I is reshaping how accounting can be done and beyond with Doctor Kan.
Welcome, Doctor Akan.
Hi, Raja. This. I'm happy to be here, Excited to be here. Awesome.
So, um, I'd like to kick it off with saying if you could, uh, share a little bit about your journey into the world of accounting.
And where did it, uh, get integrated with a I in this particular field?
What initially sparked your interest?
Sure, sure.
Um So, professionally, I worked in an adjacent field, so I worked in finance, so I worked in finance banking before moving into academia full time. um I've been in academia for for about 10 years now, About 10 years, Um, which prompted me to pursue my doctorate in accounting.
So I I've got four degrees bachelor's, master's accounting, master's in accounting finance, and then my doctorate, Um, now moving into a i it it was not, uh, uh, necessarily intentional.
So it was, uh, intentionally unintentional. I guess you could say that.
Uh, really?
When chat GP T was released in November of 2022. So that last day of November, November 30th of 2022 Like many people I I started playing with it and and just seeing how it it, it answers questions. So I decided to make a video of me plugging in auditing questions from, uh, an old CPA exam and seeing how it would do. So I recorded that video, posted that video on LinkedIn.
It caught some attention and, uh, by gentlemen, uh, of the name of David Wood, uh, from Brigham Young University. And he was working on a paper that, uh, involved over 300 accounting professors, and that would be that would make him very famous.
But it would also, you know, uh, involve 300 professors became and and it still is one of the most cited and, uh, impactful I'll say that, uh, papers on, uh, a I in, uh, in the county. So So through that I got some attention and also sparked me to just start moving forward with experimentation, writing articles and and really working with, uh, you know, a I or large language models.
Was there any, uh, recent other project that you got involved in with with A I, uh, in addition to the, uh, paper that you wrote.
Uh, I'd also like to take the link, and I'll link them in the show Notes.
What was the name of the paper that you did?
Oh, I did a couple of papers so I can plug my book, Uh, with Doctor Scott Dell, which is chat GP TNA.
I, uh, for accountants.
So, uh, we we recently, uh, published that book in I want to say it was June came out in June this year, so that was a great project, which has quite a bit of case studies. So we give you case studies, um, ways of using a I, particularly in the accounting business.
So So that's been very, uh, you know, that's been a very well received book, my book, uh, through Emerald Publishing, which is called Future Proof accounting. I talk a little bit about a I, but it's mainly about technology that was released a month later, so that was released in July. That's more of it, from an academic point of view, but it's also a very, uh, helpful uh, book.
Um, as far as research we have, Scott, Dale and I, uh, have a, uh, paper that is is being released in the next edition of the, um What is it?
Issues in online education journal in the UK. So I have another another, uh, research journal paper.
So, uh, so different projects tied around academia, Um, which which have been well received. Awesome.
What are some of the things that have surprised you when you started combining chat GP T in in the field of accounting?
Like, what is your like, You know, most surprising observation since then.
I mean, my my observation is that I, I believe this thing is gonna take some jobs. So that that was my initial observation. This is this is going to W when you would hear a lot of the media, uh, conversations.
And you, you you hear about about jobs disappearing I. I wholeheartedly believe that that was the first thing that came into mind, particularly in the, uh, accounting area.
Which type of jobs or which type of, uh, roles would be taken away in your experience?
And why do you say that?
I mean, it actually makes a lot of people nervous And, uh, like, you know, people are afraid of that.
What type of roles do you think may get affected?
And what should someone do about that?
Well, you're working on a laptop and you're you're doing some sort of, uh, work that involves text, which most people you you're writing email you, you know, if you've got to send a memo, if you've got anything dealing with generating some sort of text, it's gonna impact you. And the way it's working now is.
And and I think what people are not catching is that many people believe that, like magic, millions of jobs are gonna disappear. So it's it's not. It's not like a, uh, it's a it's a slow burn, meaning it's increasing the people who are using it. It's increasing their productivity.
Um, it's allowing them to do more with less. So if you think about someone who's not using it compared to someone who's using it and they've got maybe, uh, an increase of productivity 1015 20% that's huge if you think about it. Not not on a daily perspective, but you think about it over a a month over month perspective. They're able to do way more now. That translates to if I'm an employer.
Hey, I prefer the employee that can work at 100 and 20% versus the employee that works at 80%. And eventually I'm not gonna need the 80% employee because I can have one employee that does the work of the two. So I think that's what we're we're, uh, seeing we're in.
In my opinion, we're seeing a situation where if I'm if I'm an employer, I may have, Let's say for Exa and I can give you an example.
Uh um, of four employees.
Now, if I'm implementing a I, I've got my employees, uh, using a I productivity increased.
Well, if one of those employees leave and attrition is a natural part of business. I don't need to hire someone else. I can still use those three.
And they're they're working at 100 and 50% or 100 and 80% so I don't need to hire someone else, and that's gonna save me money, right?
And then another person leaves. I can keep going with those two people because they're doing the work of maybe nine or 10 people.
So, I, I don't need to, uh, hire anyone else. So I think we're seeing seeing that slow burn, uh, with job loss and not, uh, not not an immediate hit where companies are just firing everyone across the board. Yes. What kind of, uh, job types may get affected.
Is it like bookkeeper?
Or is it somebody is working on auditing somebody doing financial flows in the broad space of accounting, which is, uh, more prone to productivity.
Uh, right now, with activity and other A I technology, all all of them, all of them, all of them. I'll tell you, I've seen a, uh I met with the founder of a platform that's used in accounting called materia dot a. I He ran the demo of that, Uh, that's it to answer your question about all of them.
And and I asked him So this platform and I'm not a very, uh, a technical person. So So I may be, you know, explaining that I'm I'm giving you my my version of how this thing works.
Uh, because I was asking him questions. So imagine it's it's chat GP T. You can load it up so your firm you can load it up with your client is secure. You load it up with your client data, you load it up with the accounting standards or rules and regulations, and then you can go to work asking it questions, having it to to do a particular accounting, uh, task key thing.
It does not hallucinate. There is no hallucination because you're in this secure environment.
And I and I was asking him I said, Well, what is this thing run on?
What do you what is it connected to?
He said, Well, we can connect it to any model that is, uh, that runs the best. So if it's llama, if it's, uh, you know GP T four. If it's another model, we can connect it to it and run it, and that's what's running it. But it's in a secure environment, and it doesn't hallucinate. Once I saw that, I said, You know that that's it.
This this proved my thesis about the jobs. So I asked him the question.
I said, This is gonna take a lot of jobs because I mean, you can have one person who's an associate doing the work of three people.
I mean, literally.
If I mean, when he was going through it, I mean I, I said, You, you you could I mean, it was amazing. And to his point, he said, Well, there's a pipeline issue. There's a shortage of accountants So he did had a great answer for this. There's a shortage of accountants, so I don't see it taking jobs. But it would help with the shortage of accountants.
So now you got people being able to do more firms being able to work with less people again with less people and do more so a across the board it it's it's impacting uh, the profession, from bookkeepers to audit associates tax What?
What?
It it is impacting it across the board. Are there any things that you did here in your own personal life.
Uh, maybe related to accounting.
Let's say, in the recent two or three weeks, uh, which which blew your mind in terms of Oh, this is how you would have done it in a traditional way versus using chat GP T.
Uh, this is how we did it.
And wow, this is a phenomenal experience at a use case. Yeah.
I mean, yeah, I'm I'm I'm using, uh, chat GP. I'm a team's user. I'm in there every day, and it helps me, particularly with lesson planning, thinking about ways of, um, uh, approaching certain discussion topics in the classroom. I'm huge on tree of thought.
Prompting, um, I did well, I had a a webinar on chain of thought and tree of thought, uh, prompting and having the a i to do the prompt to write the prompts for you cause I'm H, you know, I. I always say I'm good at I I'm great at prompting, but I'm horrible at writing the prompts, so I know the the methodology.
So I asked the a i to write the prompt for me, you know, using uh, COT or TOT. it does it, And I get really great prompts and I get really great results. And that helps to, uh, refine a lot of what I'm doing.
Um, it's great at giving different ideas, so I may want to talk about a particular topic in class that may be very technical.
So I can use the language model to help me to, uh, explain this and I can say explain it as a undergraduate student and, you know, in college, so it can adjust the levels and gives me different ideas and ways of of transferring or or discussing a topic that may be very complex. That is hugely hugely, uh, helpful to me.
Um, some folks may not understand COT versus COT.
Can you explain?
What's the difference between COT chain of thought and three of thought?
Yeah. Yeah. So chain of thought is more you You're prompting and giving steps. So you're asking it to follow a step, you know, step one, step two. Step three, step four and giving the answer. So it's a more robust way of prompting.
So instead of really using a language model like a search engine, you know, perform a financial analysis well, chain of thought first, you know, perform a, uh, a, uh, trend analysis on the consolidated information. After performing the the trend analysis, uh, calculate the current ratio, you know, calculate debt to asset ratio, so it gives you a list of things it's going to do and then summarise your findings.
So you have a list tree of thought. Prompting is asking for different perspectives.
So you may ask the model to do those things, but then perform the analysis as a financial accountant, perform the analysis as a revenue accountant, uh, per, perform the analysis as a financial analyst and then, based on all of those analysis which pick which one gives the the, uh, most interesting or is the most, uh, technical in the results. So you'll get those different perspectives, and then you'll get one output.
So you're really you're, uh it's a way of maximising Uh uh, you know, I'll give you an analogy. I'll say. It's kind of like, uh, if you're, uh, Lewis Hamilton. He's a Formula One race car driver. You're Lewis Hamilton. You're in your, uh, F one car, and you're driving it at 15 miles an hour.
That's kind of like, if you're just typing one sentence in as a prompt by doing using chain of thought prompting or tria thought prompting or other methods of prompt. Now you're driving at your full speed.
Right?
So now you're pushing that car to its full speed and F, you know, higher level of performance.
Uh, and the thing that, uh, Roja I was mentioning before I don't write these pro.
I mean, this could be long prompts. So I'm not good at writing those prompts, and I I'm I'm not gonna try to write them. I asked the a I to write the prompts for me, so, you know, you can ask it.
Hey, this is this is what I'm working on.
Can you write a chain of thought or a tree of thought prompt for me?
The the analogy. I love it.
You know, it's going 15 MPH, but you know, you're going to pace, which is close to the full speed. Amazing.
Um, if it is gonna take away jobs, are there any ethical considerations?
If there's a role of standards, body or regulator that need to come in in terms of a looking at it from a perspective of how it has impact on the industry and the society.
What are your thoughts there?
Uh, Yvonne.
Yeah, it definitely it it's gonna have an impact on jobs. I think what's going to be the big dividing line at this moment is the pricing is the cost. And so when you look at the data on how people are using it many people So you have, uh, people, which I call self directed.
What does that mean?
You've got someone who?
Uh, for example, they may work in HR. They may work from home. They they've got their, uh, work laptop. They're working away. They've got another laptop where they're in chat GP T answering whatever questions or or or working, you know, helping them to do whatever work it is they're doing.
But they've got chat GP T somewhere, right?
S. So they're they're not using it Officially, they're using it unofficially.
So, I, I guess that's the point. And you have many of those users who may pay the the the price for the plus or the teams, you know. So they have access to more usage around a paywall. And I think the the impact of that Paywall and that pricing we're going to see more and more of of that because, for example, I'm a team's user, so I can get on chat.
I can be on there all day long, and now I have access to the new model.
You know, the open A I 001 I I've got access to it. If you you're a free user, you do not have access to it. So I think that, uh, with the paywall and the pricing that's going to keep many people from having not necessarily access, but real relevant access.
Because if I if I'm using the free version of Claude or I'm using the free version of 40, eventually, I I'm gonna type in a couple of questions. I'm I'm done. They'll tell you You timed out. So now you'll have more of a rift.
And I believe especially, um, with what I saw about the input output token pricing for the this new model, the price is gonna go up and you know it's going to make a AAA gap in in not necessarily intelligence but productivity, because the people with the money are going to be able to, uh, have access to the models for longer and be more productive than those who do not have access to the models or the high performing models.
So eventually we're gonna see tears, you know?
So, I, I think that that is the big issue around the access and, uh which which is impacted through the pricing.
What are some of the challenges that you feel organisations face in getting adopted, uh, in in inside their institution organisations?
Right, So, uh, if somebody is trying to bring a i into their accounting department and teams, what are their challenges today?
The challenge is deciding on how to implement it. What platforms to use, Uh, because there's an investment of, uh, you know, there's sweat equity, so time and training. So you've got to bring in a particular, uh, platform. You've got to train employees and the the, uh, other challenges. Things are constantly changing. So you think about, uh, chat GP T 3.5. It's gone. It doesn't It doesn't exist.
I go into my, uh, chat GP T now and you go through the models GP T four is legacy If you go in there GP T four is at the very bottom. It's legacy.
And you think of the four months ago?
Yeah, it came four months, four months ago. That's right. Yes. And it was It was the, uh, top model.
So that can be, uh, scary, you know?
And it can be a challenge because and we're not told when this is gonna happen, you know?
Or what it would look like. And a lot of this stuff is is secret.
You know, uh, so that can be challenging, uh, for, uh, organisations. It can also impact the culture, you know, because people are are concerned and don't do not like change, um, and having to learn new things so that can I it it can be an issue. It can be an issue now on the flip.
So if you're not able to overcome those challenges, you you're you're going to be, uh, really left behind compared to your competitor or another company that's a able to figure it out and able to move forward and is able to potentially stumble, right?
Or maybe fumble a little bit, uh, and moving forward with it because they'll make the the gains are are enormous or the upside is very enormous.
What are some of the trends that you think will shape up?
Uh, we just talked about how, like, within just a few months that, uh, the latest model become legacy.
How do you think?
Uh, things are looking going forward. I think going forward eventually, we will reach a plateau and, um, I.
I mean, I'm going, Raja, I'm I'm giving you my whole kinda big thought process on this.
Um, I believe if we come close to a G I or we get a GII I My personally, I don't believe that the average person will have, uh, just complete access to it.
What do I mean by that?
I think something like that would be so valuable that either you you know, if the general public would have access to it or part of it, you have to pay a tonne of money for it.
You know, in comparison to what we're paying right now. So if if we're paying $20 to access the plus version of, uh, chat GP T, whatever that will be, it will be 10 X 20 X 100 XO of that or we won't have access to it at all. We'll have access to something that is maybe 50% or or 30% of the capability of that particular model.
So I think some somewhere I, I don't think moving forward, will we?
You know that.
Hey, I, I Unless I'm paying a tonne of money, I'd be able to access. Or I would pay money and and be able to ask five questions, you know, or some something like that.
So because it's it's like you've got a genie, you know, you gotta you you're able to talk to a genie, you know?
And, uh so I think we're moving towards that. I think a G I is is coming.
Also, there's a question of what I mean, what is a G?
I So that definition I was gonna ask you that.
You know, in your view, what is a G?
I everybody seems to have their own definition of a G.
I Well, I wrote a paper, so if that that's a good one And the paper is have we already reached, uh a G I Same question.
I always ask II I mean, if you ask me, I wrote a blog post saying that, You know, we we almost reached a G I if you think it is the Turing test, but if it is a little more than that, then maybe we shifted the goal post.
Well, I think the Turing test that's a great one. the way I approached it is because I kept seeing all of these benchmarks on the undergraduate exam.
Well, OK, let's talk about the US If it's scoring, I think at the time, I don't know what it is now, but at the time, I think it was, uh, Claude and, uh, for Omni with something like 8680 88% on the undergraduate on this benchmark.
Wow, why is that amazing?
In the US, less than 50% of the population has an undergraduate degree, right?
So So it's It's around, I think, 44%. So if this thing is 88% and a minority of the population is at a college level I, I would say it's smarter than that. So if you think about that, yeah, it's smarter than most people.
The other, um aspect. So you can you can take that and look at it for, uh, graduate reasoning. It's even fewer, you know. So and then when you get to PhD, that's about 2% of the population. So and then then, uh, you go to reading level majority of Americans, uh, read around eighth grade level somewhere. It's so there's there's, uh, uh, a small percent most A larger percentage at sixth grade.
Then you get eighth grade level, you know, So you may say, Well, wait a minute.
Why is that?
Well, if you the CDC, if you think about medical labels, it has to be written at 1/8 grade level. Most people read at an eighth grade level or literacy rate. So you so if this thing is better, is is can can write as a college student.
Hey, it's smarter than than most, uh, you know people, particularly in the US. So and if you extrapolate that the undergraduate point outwards, I think in the world 10% of the world population has an undergraduate degree, some somewhere there about. So it's smarter than most people.
So do we already have it?
Is it already there?
You know that that's the question and combine what you were mentioning the Turing test.
So hey, is is it you know, do we already have it?
Uh, I think that that is a question and rod, one more point on that is I think it's important to think of it in that perspective or entertain it because, you know, I I in this space, I'm constantly reading and listening to things. And the focus is on being as smart as all PhD S on the planet.
Well, OK, but what about the other?
You know, Is it smarter than everybody else?
You know that That's a small number of individuals. So I think looking at it from both ways, um, we we may already be there. And I think that, uh, W when I listen to, you know, you listen to Sam Altman.
Uh, Dario Amadie You listen to them talk and you think of it from that perspective. I think they understand that as well, particularly when they talk about a G.
I and, uh, it's coming and and the models how smart they are or how you can apply them.
So are there any, uh, particular breakthroughs that you are expecting or excited?
Uh, that could be a big deal for more from a financial management and accounting perspective within broadly, in in the progress of a I are you seeing something that is missing that if it came, would again create like a big moment, like the chat GP D itself?
II I would say more usage. So more more companies implementing platforms and understanding platforms.
Um, particularly I. I mentioned one platform called material dot a I, uh, after seeing that II, I said to myself, That's it. Th this is that's a game changer because so for me, I'm around, uh, partners at KPMG EY Deloitte. And they're talking about their proprietary systems and how they're using them.
So learning about and meeting with the founder of, uh, material dot a I helped me because now I understand what they're building. So in other words, they've got their platforms where they can upload all of their client information. And this is thousands upon thousands of clients over years and years. And now they have a closed, secure system, no hallucination, which is key, that they can ask it questions.
They can have it perform tasks, and I mean it It it's amazing. So I think that's a huge breakthrough in, um, in the technology also understanding, uh, the difference. So there's a difference between what KPMG is using and what I'm sitting on my home laptop using. There's a big difference understanding that the other thing is that I think there's going to be more and more platforms built on a I.
And again, I'm not the technology pers you You're you're the technology person. But similar to that uh, uh, example with the material, you'll you'll have some sort of, uh, shell. And it's powered by an A I model. It's not on the A I model. So you have something where if if, uh llama is the best one to use, I can connect it to llama and run whatever it is that I'm doing.
So I think we're gonna see more and more businesses.
Um, using the A i as the the fuel or the backbone to something else that that is on top of that fun.
Um, where do you go to or what do you read to keep up?
Um, like, what kind of news or twitter handles or, like, you know, newsletters.
Uh, that you, uh, follow to keep up with a I And like and especially with the intersection of accounting. Yes.
III I get the neuron I get, uh, What is it?
Morning brew.
Um, the Journal of Accountancy.
The, uh, open a I blog.
Um, I watch podcasts. I'm I pretty much all over, like a hodgepodge of of things. What podcast do you go to do for keep up with a I, uh I listen to, uh, David Shapiro's in interest. He's in North Carolina where I'm at. I listen to him.
Um, what is, uh, the, uh, a I?
I think it's a I know how.
Uh, podcast.
I listened to that, Um, what else?
I listen to tech talks daily podcast, um, and what tends to happen, And then I follow like David Wood on LinkedIn.
So what tends to happen?
I'll get one piece of info. So let's say, for example, I'll get the neuron and they'll mention something in there, and I'll get it. It's across the board. So I get one piece of information or I'll see one thing, and then it's across the board and then I'll go to different sources. I'll listen to podcasts about it. I'll read a little bit more about it.
I'll go to So, for example, the release of the new model.
Um, that happened yesterday. I got AAA notification on LinkedIn that prompted me to go to open a I That prompted me to watch, uh, a video, you know, podcast on it to get different perspect and definitely prompted me to to to open my computer and start using it. So it's different sources different takes.
Um and I think that's important now Is is the, uh Do I like it?
No, but I had to train myself to embrace, you know, learning and really trying to keep up with all of this because it keeps changing. And it's hard. It's it's It's like, I've got another job and my other job is to keep understanding what's going on with, uh uh, a I. The other thing is, I take a lot of certificates, so I've taken the Microsoft generative a I certificate Google.
Uh, I've done two Google certificates which were focused heavily on Gemini.
Um, I'm taking quite a few certificates from IBMIBM has a a really robust training and certificate programme.
Uh, on on a i it it is tied to Watson, et cetera.
But, uh, you can you can get a lot of information insight, especially for someone who's not a technical person.
In that sense, Yvonne, as we wrap up, uh, can you recommend one key resource or strategy for, uh, professionals who are interested in using a i to improve their productivity?
You, you You said it. You gotta use it. You have to use it.
Um, I'm more of I lean towards, uh, you know, open a I and chat GP T because of it can do so much.
Uh, and what do I mean by that?
So, you know, I talked to my colleagues, and many of them say, Well, oh, you know, it just generates words.
Well, I like No, I can have it create a A spreadsheet. I can have it.
Uh, I can type in something and have it make a PowerPoint presentation. It'll download it in PowerPoint. It'll download something in word. I have it create QR codes, you know, for for presentations within seconds.
So, um, they get shocked with that, But how do I know it can do all of these things?
Well, of course I read but I use it particularly that QR code.
I didn't know it could make a QR code until I asked it, Can you make a QR code for me?
And it did it. So using it is very important because you can understand what it can do and what it can't do as it pertains to your particular workflows. And then you can add it to your workflow or what you're doing. And the only way to do that is like you mentioned using it and using it as much as possible.
And I think and we're at the time now where this is, you know, when they talk about the equaliser. So right now everybody has that it's free and as as far as access, we pretty much you're using the free version.
You have access to someone who's using the paid version for not as long, but you have that same amount of time which allows you to do develop the same types of skill sets with it and understanding with. So that's remarkable.
Will it stay like this for for forever?
I do not believe it.
Will I in, you know, at all, But you have that opportunity.
So and because the thing is changing so much that it's hard to you're not gonna be able to develop a course, you know?
So imagine you, you know, some you.
You know, you make a a university course on chat GP T 3.53 0.5 is gone. You're gonna, you know, make a course on it. So you have to keep learning it, keep learning it and keep uh, embracing, Uh, the change, uh, with it.
But yeah, definitely use it, because if you don't use it, it's It's like that race. It's kind of it's It's like, uh, you're in a race and if you're not using it at all, you know, the the starting gun goes off. Your opponent has got 15 minutes of a lead on you. So now you decide.
OK, I'm gonna get in this race. You've got to catch up.
You know, you gotta try to to to to break that 15 minute lead just to be on par with that person who's also in motion. It's gonna be hard. So putting it into your workflows, uh, is very important.
Um, because I'm I'm telling you if if you're listening to. Your colleague is using it. They're using it.
So?
So whether it's your computer, uh uh, scientist. Whether it's helping them to write code, checking that code, they're definitely trying to find a way to to do more with less and to work faster and to get things done. So they're using it. I'm using it. I know my other colleagues are using. It's there. Students are using it.
So if you're kind of stuck that can impact you and you don't understand what it is, you're kind of not understanding what's happening all around you. So I think it's That's very important. IF. That was fantastic. Thank you so much for, uh, being in the podcast. It was a great conversation.
Oh, my pleasure. My pleasure. If anybody would like to reach out or have any questions, uh, LinkedIn is a great way to connect with me. I'm always happy to help.
Yeah, I'm gonna link them.
Uh, link the coordinates to reach out with you on LinkedIn, Uh, in the show notes. Awesome. Thank you so much.
Raja, This has been great. Thank you. Great