🎉 Thanks for being a part of the CFO Circle Community! 🎉

Episode 18: Busting Artificial Intelligence Myths in CFO’s Office

Glenn Hopper_cfo_videocast_hrc Glenn Hopper

CFO

Sandline Global

_cfo_videocast_hrc
Madhurima Gupta_cfo_videocast_hrc Madhurima Gupta

Senior Product Marketing Manager

HighRadius

Available on

Synopsis:

In this episode, join Glenn Hopper, Chief Financial Officer at Sandline Global as he breaks down myths around Artificial Intelligence in CFO’s Office and why finance teams should invest in emerging technology.

Transcript:

Madhurima Gupta:
Hi, welcome to the mid-market CFO circle, a podcast powered by Highradius. Today, we are gonna talk about how and why should you be busting the artificial intelligence myths at CFO’s office by 2025. The artificial intelligence software market is expected to reach 134.8 billion dollars. This implies that organizations of all sizes should either already have a plan to leverage emerging tech or they are in process of defining one. In this episode, we will demystify the preconception surrounding AI implementation and why adopting AI is an investment and not an expense in today’s finance world. On that note. Let’s welcome our guest today, Glen. Hi Glen. How are you doing?
Glenn Hopper:
Doing good. Thanks for having me.
Madhurima Gupta:
It’s so great to have you with us. Thanks for taking the time. So before we get started, I have a quick introduction about you. So Glen Hopper is an author, energetic strategist who blends financial leadership with expertise in data science and technology. He’s a former Navy journalist filmmaker, business founder, and he has spent the past two decades helping start startups, transition to going concerns, operate at scale and prepare for funding or acquisitions he’s currently working as a CFO at sand line global. He’s also a contributor to AI journal. So Glen with such a diverse, you know, experience and career that you’ve had what is that one thing or one crazy thing that you’ve done that you’d like to share with our listeners today?
Glenn Hopper:
You caught me off guard with that. So I guess, I’ll throw this one out here as far as, things that people may not know about me. I wrote in produced an independent movie back in 2007, it’s called the hangman and it actually got distribution. It was on Netflix for a while. It was supposed to be a supernatural thriller. It was not the best movie in the world, but I basically, I was working full time when I did it, took two weeks off for my summer vacation and filmed a movie, and got distribution on it. So that’s probably the weirdest thing in my, career history that people may not know about me.
Madhurima Gupta:
Interesting. I think I’m gonna have to watch it now since we are speaking, I mean, calls for watching that movie, for sure. So thanks for sharing that. Now we’ll talk about the thing that we are here to talk about, right? So, the first question that I have for you is you know, because of pandemic and posted CFO offices are now going through severe transformations. And as far as it goes, generally offices they’re digitized, then they automate, and lastly, they power it up with artificial intelligence. And in my opinion, I think the final step of leveraging AI is really crucial for CFOs in today’s staff economy would you agree with me?
Glenn Hopper:
I do. And I would actually, I would go one step back before you can digitize. You have to go through and very analog step, very boring step, not very sexy, but you have to define your processes. And for me, even as a CFO, I want to go, I want to know when I first get to a company, I wanna go all the way to the front end of the sales funnel from leads and prospects and winning clients. And then, you know, whatever information I can get from them front and see how we’re gathering this information and whatever CRM tool we’re using. And if that information is getting passed onto another system, or if it’s just dying on the vine, once they get out of this CRM process and how the information flows all the way through. So the, the first step that I think is you’ve gotta go back and define your processes and then you know, what to digitize and what to automate. And then that’s where you’re gonna find the efficiencies is actually solving for a problem. And then I guess on the other side of that, so, you know, you define the processes you go through and digitize what you’re doing, you automate, and then adding the AI. I think a lot of times finance people in particular. I mean maybe unless you’re working in, in a world where you’re very close to machine learning and artificial intelligence, it can be intimidating. So I think that what people are missing out on is wider world of artificial intelligence in the smaller world of machine learning and even things like robotic process automation. So if people think that machine learning and, and artificial intelligence mean that you have to have the, how computer from 2001, or, you know, some kind of sci-fi version of AI, it’s not that there are smaller steps you can take. And there are levels of artificial intelligence that you can do without a great deal of data. So robotic process automation that’s to me is a good first step in, incorporating artificial intelligence because you’re basically with RPA, you’re training a computer, the same way that you train an employee, you’re saying, Hey, we’re gonna use robotic process automation to bring these invoices in and enter ’em into the system. So you’ll train the software program, look in this section of the invoice for the address, look in this section of the invoice for the vendor name, look at this section for the amount, and then you can train it if an invoice comes in from this vendor and has these words you know, code it to this account. So that’s some pretty basic stuff. , and that’s, that’s not very daunting for people if you have the software and, and you have it in place to do RPA that’s a good first step, but I think people need to think about, okay, there’s that first baby step, but then what can we do. They need to see the end goal and understand incrementally how to, that there are smaller steps to, to get there. So I think, you know, maybe RPA is the first thing you do. And then the next thing I’d say with machine learning, you can get cognitive insights on what your business is doing by cognitive insights. I mean, like think of it as analytics on steroids. So based on the data that you have. So, like I said, if you go all the way back and get information from the beginning of the sales funnel, you can predict based on as much, you know, as much data as you can get early on what customer is likely to buy or not buy, or maybe on the other side, you can see these are commonalities and correlations that we see before a customer turns away. It’s taking stock of the data you have and using that to get insight into your business. And that in order to do that, you have to collect the data. You have to comb through it and transform it and, and make predictions on it. I guess another way to describe it is you do the RPA and so you, and once you start gathering that data, you can have descriptive statistics. So you basically, here’s all the data we have. Here’s what we know about our customers. And then once you do that, you can start predicting what future customers may or may not do on the data. So that’s where you’re getting the insight and then cognitive engagement, which is, is kind of the, the top level of what you’re doing with artificial intelligence. That means you go from descriptive to predictive, to prescriptive, meaning you’re actually taking this data and making action on it. So if you see, Hey, these customers are going through something that’s very similar to customers who turned away, what is something we can do to stop them from churn. So kind of a long-winded answer, but I think that it’s so important that people understand it’s not, you, you don’t just jp in and suddenly have robots running your company. You, you can take these incremental steps. And as, as you move down the pipeline with artificial intelligence, you can get increased value out of it.
Madhurima Gupta:
So even, at HighRadius, we talk to a lot of, CFOs. And one thing that often comes across is hesitancy to adapt to an AI based solution because they do not have their data, right. And unless they have their data, right there isn’t any, predictive or prescriptive data modeling, then that can be used on top of it to actually make data-driven decisions automatically. So I think that is one of the key, I’d say hiccups that CFO offices do come across, but you know what you pointed out that they can first start using RPA and get that data in place. And eventually, you know, you know, proceed to the next level of automation. I think that that’s a great way to do that. You know, so the scan also help you improve your data governance strategy. And, and what would you say, will they be missing if they don’t adopt AI, right. The last step that we talked about,
Glenn Hopper:
I mean, digital transformation is hard. I’ve been in this space for really the last, I’ll say a little over five years of helping companies, small, you know, the 10 to 20 million revenue companies really get to the place where they can compete with the larger companies and they have limited resources. And if you don’t have a handle on your data, I mean, it, it can hurt you in a lot of ways. One is if you ever want to raise money from private equity, or if you ever want to, maybe participate in M and a be sold and you don’t have a good handle on your, your data, then you, are not gonna be as valuable to investors or acquirers. , you’re also not gonna be as efficient, but when the real part that they’re missing out is those last two steps of the predictive and prescriptive, tools that you can use and really drive value to the business. So historically finance has been a report card, talking about something that happened after the fact, and as more and more of the finance function is automated. , the value of the finance department, if you’re not looking ahead, and if you’re not embracing new technology, then finance is often seen as a cost center to companies. And if all you’re doing is reporting historical information, then yeah, you are a cost center at this point because everything you’re doing can be replaced by software. So you’re really missing out on this chance to actually add value to the business with this analytics from the finance side.
Madhurima Gupta:
So Glen, moving on, we actually came across a recent survey wherein, finance leaders expressed great concern over, lagging, considerably behind their, counterparts and supply chain, HR product departments, when it comes to adoption of AI capabilities. Why do you think finance leaders are falling behind others when it comes to AI acceptance and adoption?
Glenn Hopper:
I think one of the issues that those of us who’ve been in finance for a long time have come across is it’s what I’ve referenced in the last question where we, we can be seen as a cost center to the company it’s front of the house stuff, product stuff, that’s the sexy stuff. That’s what companies love putting their money into. And it’s, you can immediately see, Hey, we’re developing this new product, this new feature, it helps grow the top line and it’s out there and it’s newsworthy. , whereas the back of the house, it’s, it all just seems like, what’s that expression, how the sausage is made. Like everybody just wants to see the sausage, not what goes on and how it’s made, and it can be cumbersome and clunky in what we’re doing in the back. And it really, if you are trying to find budget for, artificial intelligence and you’re competing against those front of the house, things that people understand more versus, you know, trying to explain RPA to someone who, who doesn’t know what it is, or, you know, artificial intelligence and machine learning and how you need it. If you are seen as this cost center and you are asking for more money to do what you do, and you don’t, you can’t speak to what we just said in that first question of how you’re gonna add value, then it, you are gonna get left behind because if you, and thinking about like return on investment for, investing in this technology, it can be, if it’s considered an expense, it can be, the company thinks, well, you’re crazy. We’re already spending this much on our back-office operations. Now you’re asking us to spend even more, you know, we don’t have it. We need to put that, that money elsewhere. So I think it’s primarily what people see the most is what’s gonna get the most attention and because kind of this stuff that happens in the finance department in the back office like nobody even thinks about it until you make a mistake
Madhurima Gupta:
Talking about expense, Glen, I’d come to my next question. So in general, the notion that exists in the market is that the implementation cost for AI is one of the most expensive things to implement in the first place, which often becomes a barrier for its adoption. So, what should, what would your recommendation be for, different CFOs in terms of, should they be looking at implementation or investment in AI as an expense or something else?
Glenn Hopper:
ROI, if you were just talking about, let me invest this money and let me show you an immediate return on it. It’s, you know, if we buy this computer package, we can replace an employee or, you know, whatever you’re, you’re saying, if you’re looking for an efficiency, but at the same time, I, I think you worded it correctly in your question, in that if you look at it as an expense, it can be hard to justify an ROI. But if you look at it as an investment, like if you’re building widgets and you’re, you reach sort of maxim capacity of how many widgets you can build in a certain factory, well, you to grow, you have to make the capital investment in a new factory. So nobody says, oh, don’t make, don’t spend that money. Just stop where you are. , and so the, you know, the return on investment for that is we’ve gone from one level where we could produce a million widgets a year to now we’ve invested in this new factory. Now we can produce 2 million widgets a year, and then you, and that has ripple effects all the way through. So if you don’t look at it like, oh, we are adding, another software package, another bit of technology, maybe some upskilled employees who are a little more expensive than prior employees. If you just look at it as finance is a cost center, and now finance is telling us, they’re gonna cost more. If you don’t understand the value, and if you don’t understand the investment, then you’re having to fight the fight of, okay, you got robotic process automation. You said you have a software program that can input invoices now. So that means, you can get rid of an AP clerk and maybe I can get rid of an AP clerk, but when I’m making the argent for it, I’m not saying I wanna get rid of anybody. I’m saying that person who was an AP clerk, I want to upskill them and make them a data scientist. And you know, if it’s not, if the person who was doing AP clerk, isn’t the right person. Then you may have to replace them with someone with different skills. But every time I talk about investing in AI and ML, it is, this is going to cost you more. And if you’re looking just on the income statement at what the return on investment is gonna be, you’re not gonna find it there where you’re gonna find it is we are making, we are helping you with this data to make more strategic decisions that are gonna help you grow the top line and are, if you can find the magic formula that predicts when a customer’s gonna churn, and you can stop some percentage of those customers from churning. If you think about your customer acquisition cost, that’s more valuable than new customers a lot, because it’s easier to, or it’s, it’s cheaper to maintain the customers you have than to bring new customers in. , conversely, if you are using machine learning to predict what’s going on in the market, what’s going on with your products and you can make better decisions about where to focus your sales efforts, what channels, what markets, what regions, and all that’s driven by the data that you are using in your back office. That’s where the value comes from. I love the finance field. I’ve been doing it for 20 years, and I’ve firmly believed that we are the ones who should be leading analytics because we’ve been doing, polynomial regressions in Excel for 20 years. And we inherently know how to do this. We just have, it’s incumbent on us as finance professionals to learn the new, new technology and to fight this turf for and show that the finance role is evolving. And we’re not just historical reporters. We are the keepers of this data who can help become strategic for the companies.
Madhurima Gupta:
So moving to my next question, Glenn, so we’ve talked about investments, we’ve talked about, talked about the cost of AI and the impact that it can bring to CFO’s office now, finances, definitely one of the most critical departments in an organization. , but, it can be difficult to entrust it to relatively budding technology. There are issues of ethics that come into picture, especially specifically when you’re using AI, because it can have biases. , what would you say, or would you say that it is important for CFOs office to understand that it’s not gonna be humans or machines making the decisions and mostly machines when it comes to AI, but it’s gonna be a combination of humans and machines.
Glenn Hopper:
This is a popular conversation now, and I think I actually wrote an essay for it. And I’ll, I’ll share the link with you. , I wrote an essay for it for a Harvard conference a couple years ago about, sort of the next phase of technology is going to be human-machine collaboration. It’s sort of augmented brain power. And that’s, you know, whether it’s for individuals using something, you know, some kind of AR whatever, you know, tool that you’re using that is more, you know, easier to use even than, than cell phones, or if it’s companies using technology to make better decisions. And I call it, from hunch to hypothesis. So if you’ve been in the business world for a long time, and you have experience, you , you have been trained on your experience and your intuition right now, can’t be replaced by artificial intelligence. What we have to think about is we’re automating so many things right now, and it used to be the first or the last revolution was probably robot robotics and manufacturing, and some jobs were, were lost then. And people there’s a lot of doomsday prognosticators, and sort of the Luddites every time there’s a technological change. There are people that say, oh, this is the end. There’s nothing for people to do, but historically we’ve found new things to do over time. Of course, as technology grows further and further, it replaces more and more human tasks. So, we have to be conscious of that, but I do think we’re replacing things right now. We’re replacing mindless jobs, invoice entering, paying bills, bank reconciliations, just low-hanging fruit of stuff. That if you go to school to get a finance or accounting degree, you don’t envision getting out of college and going and being a data entry person. So, and then, you know, you don’t wanna be just entering data all day. So if that can be replaced by software, you need to find that gap of, okay, now the easy stuff is done. How can we use our han horsepower to, to make the business better? So I think it’s about we’ve potentially have access to more data than we ever have. And if you can automate the, the processes and focus more on, on cleaning and collecting the data and how to use that data to drive decisions that it, it can work together. , but in order to do that, there’s a level of trust. It’s I just had this conversation yesterday, a friend of mine has a, a Tesla, and he’s got the full self-driving feature on it. And I guess that’s not a, a standard package, but we were talking about, he had, let me drive this car on the full auto mode.And I would not trust self-driving. And, you know, he would say, you don’t have to, you don’t have to put your, foot on the brake at this four way stop. It’s gonna know when to stop, and it’s gonna know when to go and I could not make myself turn it over to the car. , that said, I’m completely fine, turning over a bank reconciliation to software . And it’s, I guess it’s you know, where the stakes are, in that, but, I think you have to have a leg level of trust. And then there’s one more component in there that I think makes people hesitant. There’s if you look at the different ways that people get power, there’s sort of the power of your position, the power of your abilities, the power of the information, you have your social power, all these ways that people get power. And there is one that is, if you are the only person in an organization or in a group who knows how to do one thing, then that’s your power. And you feel like that’s your protection. And the idea of turning that over to a machine is it’s hard on the ego. So I think people have to get that to that place where they trust it and they understand it, and then they can turn it over you know, more of the stuff and, and work together. But it’s, you have to build, you have to build trust and you have, and in order to build trust, you have to understand it. And I think that maybe especially for as old heads, who’ve been doing this a long time, it’s hard, you know, that another cliche, it’s hard to teach an old dog new tricks, and it’s just, we get set in our ways and change is hard. And so the idea of trusting what’s going on to a new software system is, is very hard, especially if you don’t understand the technology, but the more you understand the technology, the more the trust is there. And I think that’s just inevitably, we’re gonna get there because people who don’t make that move and don’t see humans and machines working together, they’re gonna be the ones who are squeezed out over time. And that’s probably a pretty short timeline that they’re gonna be squeezed out into.
Madhurima Gupta:
It would. So everybody listening in should definitely move ahead off the curve and see what new they should do. , on that note what I also wanna ask you is about the biases though. We did touch upon it slightly before in the interview. , but what I wanted to understand is that one of the concerns that CFOs directors, VPs of finance, different companies have with regards to using an AI-based system, is the biases that it may have, depending on the data sets available what is your opinion on it
Glenn Hopper:
Biases an inherent problem with machine learning. And it is whatever model you train is only as good as the data that you feed it. So, you know, you go through the iterative process of continually training the model and making it better. But if you have, there’s countless examples of even big companies who’ve gotten this wrong. I think of what was a few years ago, Amazon and their, they had the AI that was selecting candidates for hiring and it, so they trained that AI on existing employees. Well guess what, if you don’t have a lot of diversity in your employment base and you don’t have you know, if it’s looking for things based on the employees who are there, and if you have a, a mostly male or, or mostly white or whatever, you know, whatever your workforce is if you’re training that on who’s there, then it’s gonna be biased against minorities or women, or, you know, whatever the case is because it was trained, Hey, this data set that we have here, these are who the great employees are. So you can’t expect the machine to see through that bias if you only feed it data. That’s like, so there’s the hiring example. There’s you’ve seen it in, in loaning industries where maybe in a, in a, a lower economic neighborhood zip code there may not be as many loans. So if you train a machine-learning algorithm to select who gets a loan, and in that data set, there’s not many people, or they’re maybe in the lower economic area, maybe there’s people with worse credit ratings or whatever. But if it just because you’re in a zip code denies you alone, then that’s bias and it’s and it either, there’s just, there’s so many examples of how you can do it. And I think that the danger and, you know, you have to think about if you’re gonna be using machine learning, you have to understand what you’re doing. I think I used this example the other day. There’s a lot of tools out there that and they’re gonna be more popular in coming years where you don’t have to be a data engineer. There’s gonna be more and more drag and drop low code, no code artificial intelligence. But if you’re not, if you don’t have a stats background, if you don’t have a data science background and you don’t understand what’s going in there, you can, you’ll just load it up with the data you have that may not be cleaned, may not be accurate. And if you don’t think about your data, the same way that a statistician would think about designing a, a survey and trying to come up with a true impartial, random sample, then you’re gonna just, you’re gonna enhance the problem. So, Matt, the biggest, the problem with bias is we’re talk if we’re talking about mid-market companies and what people are doing, you have people who are stretched thin as it is. And you know, they’re doing what they could be the controller, and they’re just focused on the controller duty. So they could be the operations guy just focused on that. If you don’t have a true data science person on staff, and you’re dealing with all this data and you don’t have someone thinking correctly about it, then you could be, it’s like, just like the flawed Amazon hiring process. If you don’t understand your data and what you’re feeding into a machine, if you’re trying to predict when a customer’s gonna churn and you don’t have the right information and you’re not looking at the right variables, then it’s gonna be probably worse than if you did nothing. So I think God, I just, I can’t stop throwing out cliches today for some reason, but I, I thought of that Spider-Man quote that is, you know, with great power comes great responsibility and machine learning is a great power. And if you’re going to use it, you have to responsibly think about the data and what the implications are for it.
Madhurima Gupta:
I think that’s a very good cliche to share on that example. , so Glen, we come to the, you know, conclusion of this particular podcast. And before we hop off, there’s one last question that I have for you think of it as a parting note that you might wanna leave with our listeners. So what is that one biggest artificial intelligence-related myth that CFO’s office needs to burst today?
Glenn Hopper:
If people don’t understand what artificial intelligence is, they don’t, and we addressed it earlier. They don’t see it as incremental steps if they just think we’re jumping from using spreadsheets. And I don’t know, QuickBooks and whatever, you know, some, maybe some sequel database stuff. If, if you just think we’re jumping from the way we’ve done things for the last decade into now, we’re on the bridge of the Star Trek enterprise you know, in, it seems like too big of a chasm to cross. So I guess the biggest thing that I would like for CFOs and, and finance people to think about is you don’t have to do it all at once. Find those incremental ways to use data. And I think that you know, small, mid-market businesses, we’re never gonna have the amount of data that like an Amazon or a Google or whoever would have. But if you look at, I’m thinking of like Kaggle contests, like the one that housing prices in aims, Iowa, you know, you get certain number of features and maybe there’s, you know, 15 or 20 different features you can use, but in your company, you could have 15 or 20 data points and you can build, if you have the right skill set, you can build a model on those data points. So you may not have, you know, terabytes of data, but even with the small amount that you have, you can start using, that data to make data-driven decisions and make incremental steps towards adopting AI.
Madhurima Gupta:
Thank you for your, suggestions and your opinions today, and for sharing it with all of us. , I really enjoyed having this conversation with you, Glen, and I hope to have you back on CFO circle sometime soon. And in the meantime, all our listeners out there, thanks for joining in and listening in, stay tuned. We’ll be back with.