Research from Goizueta’s Cassandra Estep shows decisions can depend on AI use by companies and auditors.
Artificial intelligence (AI) is developing into an amazing tool to help humans across multiple fields, including medicine and research, and much of that work is happening at Emory University’s Goizueta Business School.
Financial reporting and auditing are both areas where AI can have a significant impact as companies and audit firms are rapidly adopting the use of such technology. But are financial managers willing to rely on the results of AI-generated information? In the context of audit adjustments, it depends on whether their company uses AI as well.
Willing to Rely on AI?
Cassandra Estep, assistant professor of accounting at Goizueta Business School, and her co-authors have a forthcoming study looking at financial managers’ perceptions of the use of AI, both within their companies and by their auditors. Research had already been done on how financial auditors react to using AI for evaluating complex financial reporting. That got Estep and her co-authors thinking there’s more to the story.
“A big, important part of the financial reporting and auditing process is the managers within the companies being audited. We were interested in thinking about how they react to the use of AI by their auditors,” Estep says. “But then we also started thinking about what companies are investing in AI as well. That joint influence of the use of AI, both within the companies and by the auditors that are auditing the financials of those companies, is where it all started.”
The Methodology
Estep and her co-authors conducted a survey and experiment with senior-level financial managers with titles like CEO, CFO, or Controller – the people responsible for making financial reporting decisions within companies. The survey included questions to understand how companies are using AI. It also included open-ended questions designed to identify key themes about financial managers’ perceptions of AI use by their companies and their auditors.
In the experiment, participants completed a hypothetical case in which they were asked about their willingness to record a downward adjustment to the fair value of a patent proposed by their auditors. The scenarios varied across randomly assigned conditions as to whether the auditor did or not did not use AI in coming up with the proposed valuation and adjustment, and whether their company did or did not use AI in generating their estimated value of the patent. When both the auditor and the company used AI, participants were willing to record a larger adjustment amount, i.e., decrease the value of the patent more. The authors find that these results are driven by increased perceptions of accuracy.
It’s not necessarily a comfort thing, but a signal from the company that this is an acceptable way to do things, and it actually caused them to perceive the auditors’ information as more accurate and of higher quality.
Cassandra Estep, assistant professor of accounting
“Essentially, they viewed the auditors’ recommendation for adjusting the numbers to be more accurate and of higher quality, and so they were more willing to accept the audit adjustment,” Estep says.
Making Financial Reporting More Efficient
Financial reporting is a critical process in any business. Companies and investors need timely and accurate information to make important decisions. With the added element of AI, financial reporting processes can include more external data.
We touched on the idea that these tools can hopefully process a lot more information and data. For example, we’ve seen auditors and managers talk about using outside information.
Cassandra Estep
“Auditors might be able to use customer reviews and feedback as one of the inputs to deciding how much warranty expense the company should be estimating. And is that amount reasonable? The idea is that if customers are complaining, there could be some problem with the products.”
Adding data to analytical processes, when done by humans alone, adds a significant amount of time to the calculations. Research from the European Spreadsheets Risks Interest Group says that more than 90% of all financial spreadsheets contain at least one error. Some forms of AI can process hundreds of thousands of calculations overnight, typically with fewer errors. In short, it can be more efficient.
Efficiency was brought up a lot in our survey, the idea that things could be done faster with AI.
Cassandra Estep
“We also asked the managers about their perspective on the audit side, and they did hope that audit fees would go down, because auditors would be able to do things more quickly and efficiently as well,” Estep says. “But the flip side of that is that using AI could also raise more questions and more issues that have to be investigated. There’s also the potential for more work.”
The Fear of Being Replaced
The fear of being replaced is a more or less universal worry for anyone whose industry is beginning to adopt the use of AI in some form. While the respondents in Estep’s survey looked forward to more efficient and effective handling of complex financial reporting by AI, they also emphasized the need to keep the human element involved in any decisions made using AI.
What we were slightly surprised about was the positive reactions that the managers had in our survey. While some thought the use of AI was inevitable, there’s this idea that it can make things better.
Cassandra Estep
“But there’s still a little bit of trepidation,” Estep says. “One of the key themes that came up was yes, we need to use these tools. We should take advantage of them to improve the quality and the efficiency with which we do things. But we also need to keep that human element. At the end of the day, humans need to be responsible. Humans need to be making the decisions.”
A Positive Outlook
The benefits of AI were clear to the survey participants. They recognized it as a positive trend, whether or not it was currently used in their financial reporting. If they weren’t regularly using AI, they expected to be using it soon.
I think one of the most interesting things to us about this paper is this idea that AI can be embraced. Companies and auditors are still somewhat in their infancy of figuring out how to use it, but big investments are being made.
Cassandra Estep
“And then, again, there’s the fact that our experiment also shows a situation where managers were willing to accept the auditors’ proposed adjustments. This arguably goes against their incentives as management to keep the numbers more positive or optimistic,” Estep continues. “The auditors are serving that role of helping managers provide more reliable financial information, and that can be viewed as a positive outcome.”
“There’s still some hesitation. We’re still figuring out these tools. We see examples all the time of where AI has messed up, or put together false information. But I think the positive sentiment across our survey participants, and then also the results of our experiment, reinforce the idea that AI can be a good thing and that it can be embraced. Even in a setting like financial reporting and auditing, where there can be fear of job replacement, the focus on the human-technology interaction can hopefully lead to improved situations.”
Goizueta faculty apply their expertise and knowledge to solving problems that society—and the world—face. Learn more about faculty research at Goizueta.
