By Elizabeth M. Carroll, CPA

The question that seems to come to everyone’s mind when thinking about artificial intelligence (AI) is whether AI is going to replace their job? With regard to public accounting, the answer is probably no. However, this may not be for the reason you might think. The industry has been struggling with a staffing shortage, causing more routine work to filter up to be done by more experienced employees. While AI might not be coming for a senior’s or a manager’s job, it may take on some of the responsibilities that had been performed by staff, who are increasingly hard to find in quantity.

If you find yourself in a similar situation, don’t just jump right in. When looking at implementing any kind of new technology, it's important to consider where inefficiencies exist and how technology can assist. This applies to AI technology as well. The first step a company should take when considering automating some functions with AI is to identify the mundane or repetitive tasks the staff is doing—essentially, the tasks that are so boring it could explain why there’s a staffing shortage in the first place. These are also usually the activities that consume time and resources, draining the budget. Yes, they are functions that are necessary for overall engagement completion, but they do little to enhance staff’s experiences or knowledge base.

In audit, this could be document matching – agreeing information from a sales listing or accounts receivable report to an invoice. Back in the day, this may have involved physically sorting through dusty boxes of invoices at the client’s office. These days, it might mean getting sent one extremely large PDF file, or a folder of smaller files. If the staff is lucky, the files are bookmarked with a descriptive name indicating which PDF page goes with which selection, but often the staff must determine if the file named “Scan20230722_0012” contains one of the 60 selections they are testing.

Staff can spend hours scrolling through PDF documents of invoices for each selection made, matching information from the report to the PDF page based on certain parameters, such as customer name, invoice number/date, and amount. With machine learning and natural language processing, software with AI capabilities could scan through the PDF and identify information based on a list of selections and identified parameters, informing staff in minutes with what matches and what doesn’t.

On the tax side, AI technology could be used to sort through client documentation, but instead of matching information to a predetermined list/sample as with an audit, AI could extract the numbers out of the documents and import them into the desired location (e.g., workpaper, software, etc.), assisting with data entry. This technology currently exists, but it is somewhat restricted. Often it "reads" certain types of support, filling in some of the numbers needed but not all. If the technology were learning from each type of document scanned into it, it could begin to recognize more types of support and how to populate information accordingly.

Pure reliance on these types of technologies is mostly hypothetical, but they are out there to an extent in raw form and getting better. With any new technology, though, early adoption is not always the best policy. AI is so new, entrusting it with even mundane, repetitive tasks likely will not yield the desired results. There are also other reasons to hesitate, such as ensuring staff gain the same knowledge from reviewing AI’s work as they did from performing the work themselves. As staff ascend the firm hierarchy, they still need to understand what a sales invoice looks like or how numbers from an investment statement flow onto a return.

Implementing AI in public accounting does have the potential to streamline operations and reduce the burden of repetitive tasks, yet careful consideration must be given to ensure that technological advancements do not stunt the development of staff’s skills and knowledge base. However, as the industry evolves, AI may become a strategic enhancement that assists engagement teams rather than replacing them. Thus, AI will play a part in maintaining the high standards of service and expertise that clients expect.

Elizabeth M. Carroll, CPA, is director, research and development, with Kreischer Miller in Horsham, Pa. She can be reached at ecarroll@kmco.com.

Sign up for PICPA's weekly professional and technical updates by completing this form.

^{Statements of fact and opinion are the authors’ responsibility alone and do not imply an opinion on the part of the PICPA's officers or members. The information contained herein does not constitute accounting, legal, or professional advice. For actionable advice, you must engage or consult with a qualified professional.}