Ep. 7: Mark Nickerson - Risks and Rewards of AI

Mark's Article: https://sfmagazine.com/post-entry/april-2019-ai-new-risks-and-rewards/

FULL EPISODE TRANSCRIPT

Music: (00:00)

Adam: (00:00)

Welcome to count me in IMA's podcast about all things affecting the accounting and finance world. I'm Adam Larsen and with me is my co-host Mitch Roshong. Our topic for this week's episode is actually based on an article written about artificial intelligence. Mitch, can you tell us a little bit about our speaker and what the article says?

Mitch: (00:18)

Yes, absolutely. So I recently came across an article titled AI new risks and rewards written by Mark Nickerson. Mark is a CMA, CPA and MBA and is a lecturer for the state university of New York at Fredonia. I was interested in learning more about where the idea for the article came from and if Mark had any other insight into what kind of risks accounting and finance professionals should be aware of. He was nice enough to answer a couple questions for us. So let's listen to how AI could potentially increase finance fraud in the future.

Mitch: (00:52)

Can you tell us what led you to researching and writing the article, AI new risks and rewards?

Mark: (00:58)

Yes. Thanks so much for having me. I wrote the article in response to a number of stories that I had seen recently, all of which seem to focus on the positives of utilizing AI, artificial intelligence or RPA, robotic process automation in both the finance and accounting industries. However, there was very little, if any credence given to what the potential negatives or drawbacks of instituting these technologies would be. Now,i'm not adverse to new technologies. I'm not adverse to being on the cutting edge of utilizing or implementing new technologies. Quite the opposite. However, I always want to make sure to play devil's advocate as well as take a realistic look at what the potential negative impacts are. I think it's always important in our industry to be proactive to what could go wrong instead of being reactive as we've been in the past now in these articles that speaks quite often about how corporations and businesses will use and implement and have already implemented AI in their respective finance departments or their internal audit departments and I believe wholeheartedly that those companies are seeing a large benefit. I cited in my article a recent MIT Sloan management review report where they surveyed 3000 executives and over 85% that they felt AI would give them a competitive advantage over those companies that were not using AI. And of those individuals, 79% felt it would increase productivity. I believe that they're correct. I also believe though that it opens up a brand new world for potential fraudulent transactions, potential accounting and fraud scandals that could go undetected for years or never be detected based upon the reliance of AI to find these things. And what I mean, there circles back to what AI really is in artificial intelligence. While I'm not a technology expert, I can break it down and have had it explained to me many times in a simple format of artificial intelligence does learn, but it doesn't learn on its own. It learns from you and I, it learns from millions and millions of data points that it is fed through in analyzing information. Those data points come from situations and occurrences that have taken place where humans have been interacting in some facet of the data that we are then feeding through the artificial intelligence because humans have interacted or touched some portion of that data point or that transaction. Human bias is inherently being fed into these technologies, into the artificial intelligence, into these quote unquote robots. And I cite a couple of studies in my article, one of which took place in the university of Virginia where a professor down there was utilizing AI and determined that it began exhibiting sexist views of women. simply again, based upon the data points containing human bias and human bias. Unfortunately, still today being skewed towards towards sexist views towards a female gender. Another study this time at the University of Massachusetts showed that a AI was able to put learn in quotes again because it does take it's lessons or it's education from human interactions. But AI was able to learn from these data points to essentially exclude African American individuals from data sets that were used for important items such as polling information, based upon their vernacular. So it was able to learn and essentially pull those data points out of the data sets. My concern then is the human bias that led to the large accounting and financial frauds in the late nineties and early two thousands such as greed and power and seeking to meet quarterly goals and sales records. Those issues will still be contained within data points, those biases that have been exhibited by humans for hundreds of thousands of years will essentially be contained in the AI and how are we as auditors, as accountants, as financial professionals, ready to combat that? I think that anybody who sees the articles that have come out on how AI has taken on these previously unprogrammed racist and sexist and biased views and does not think that items such as greed are essentially inherently contained in these transactions as well are not focused on the potential impacts. What if AI does identify that greed or that meeting quarterly reports, meeting quarterly goals increasing market share, taking home more money? What if AI learns that those are all things that certain humans that are involved in data points see as beneficial and therefore AI derives the fact that it should do more of that. If it does, we could be an appoint where essentially AI begins committing accounting and financial frauds on its own if not even more concerning committing these frauds in a complicit manner with executives or individuals that have to programming those technologies. So I think the main result is we need to make sure that we're not losing the critical eye of the auditor or the critical eye of the financial professional. The executives that was an are inherent in Sarbanes-Oxley just to be on the cutting edge of technology or increase efficiencies to cut cost. That in my opinion, is the real danger.

Mitch: (09:30)

What do you think accounting and finance professionals should do moving forward?

Mark: (09:34)

Yeah, great question. So I think the issue becomes, again, that I am in no way, shape or form against utilizing AI or RPA or data analytics to increase efficiency. But we need to realize that those technologies are limited and we is the so called gatekeepers of our profession need to ensure that our reliance is still on our professional judgment and our interactions and our decision making because another point needs to be made that AI isn't as flexible as some articles and individuals make it out to be. Again, these data points need to be repetitive in nature and very numerous for AI to be able to make good quality decisions and process information accurately. So if there are situations in audit, in business, in finance that are not repetitive in nature or independent, we need to make sure that we as CPAs, and professionals are looking at all transactions to determine, Hey, that needs to be pulled out and that needs to be something that we make the decision on because it's not a frequent transaction or it's a little bit skewed from the norm, so we might not get a good result from the technolog...