Artificial intelligence (AI) promises to revolutionize accounting as we know it, and eventually—if it hasn’t already—the technology will affect your organization, too. AI and its related technologies have the potential to automate processes, and sift through volumes of data for abnormalities more efficiently and effectively than humans can. That feature creates the potential for AI technologies to reshape the world of accounting—including bookkeeping, financial statement audits, and reviews—the same way it is modernizing industries ranging from manufacturing assembly lines to telecommunication call centers.
It’s All about Machine Learning Right Now
First, a little nomenclature clarification. Popularized by IBM’s Watson, AI is the technology that can reason and learn from a task. AI processes mimic how a human would respond to an issue as evidenced by AI's work behind voice recognition platforms on Google, Apple, and Amazon devices. But AI can also be used to describe a family of technologies that have the potential to make improvements to future processes based on the results from past processes.
Machine learning is one such member of the AI family, and it’s the technology you’re most likely to encounter in accounting technology today. Like the more sophisticated forms of AI, machine learning becomes more effective the more it is put to use. It is not an out-of-the-box, fully automated technology. Maximizing the value of machine learning takes some careful configuration and a whole lot of data. Most AI developers are currently looking for ways to build significant volumes of data-mineable information to assist with the machine learning process.
The technology helps analyze large data sets because it can recognize certain elements—such as dates— across standard templates no matter what type of input (e.g., contracts, invoices) is included in the data. Machine learning is being developed that allows inconsistent data to be recognizable. In the interim, individuals contribute to the process by indicating to the machine learning program which data markers the technology should be flagging. There also needs to be some consistency around the inputs into the machine learning system—such as the use of standard templates. Because the technology is still being developed around recognizing inconsistent data, a human review element is required to help machines “learn” from unique formats. Machine learning analysis will likely not be 100% accurate until that piece of the puzzle is addressed, if ever.
Individuals are helping machine learning technology reach that point by “teaching” the machines about possible variations in the data. After reviewing the results, humans determine if the machine was correct by either validating or correcting the machine’s analysis. From there, the machine uses what it has learned from the human in a new dataset to become more accurate. This is why such large volumes of data are needed, and why accurate, manual review processes are so important in the early stages of adoption.
What Machine Learning Looks Like in Accounting Practice Today
Machine learning could help with a number of financial review processes, such as reviewing contracts for ASC Topic 606 revenue recognition or leasing. Accountants could indicate the characteristics of contracts that qualify for over-time recognition into a machine-learning tool, and the tool could take that information and identify when revenue should be recognized. The accountant could analyze the results, and help the machine learn from what it missed. After enough repetition of machine analysis followed by accountant analysis, the machine would learn enough to be able to scan through contracts and indicate how the contracts should be recognized under the Topic 606 revenue recognition rules with little to no human involvement.
The technology could also be put to work for the lease accounting changes in ASC Topic 842. Companies will need to review all of their leases for compliance with the changes in the accounting rules, and machine learning has the potential to take on some of the manual labor involved.
Machine learning appears in financial statement audits as well. Accountants can configure their machine learning technologies to help identify higher risk transactions. Machine learning allows accountants to review a greater volume of transactions; generally, auditors review a statistical sample of representative contracts, but by utilizing AI, the entire population can be analyzed faster than an individual auditor could review a sample.
Technologies can also be used for benchmarking purposes. Accounting technology paired with machine learning may process a broad pool of anonymous financial data that can help companies compare themselves to their peers. For example, accounting technology could help a manufacturer in California with $8-10 million in revenue compare its financial results to anonymous financial results from manufacturers in the West with similar characteristics.
The Promise of Machine Learning for the Accounting Practice of Tomorrow
Because of the amounts of data input required, most machine learning accounting technologies are not accurate enough to be more efficient than a human accountant, but they have the potential to be more accurate and effective in just a few years. When the technology is appropriately configured, it can reduce the risk of human error in the contract review process, and make the process more efficient and effective.
Many of the software solutions that incorporate machine learning come with a premium cost because it is still an emerging technology. The current cost of machine learning will keep its incorporation into day-to-day business functions to a modest level, but the price point is expected to decline as the technology advances.
One of the ways the technology could advance would be by providing benchmarking data in real time. Dashboards could be configured to provide real-time insights that would help companies make more immediate course corrections and other adjustments to their operating strategies.
Another evolution that is not so far off in the future will be using machine learning technologies to connect to back-office accounting functions, so the process between recording the financial information and analyzing it will be more seamless.
Investment Now, Return to Come
As the common expression goes, the best way to predict the future is to make it. By taking the time with machine learning now, the accounting function can help position itself for the future. It’s also expected to be an investment in time and money that eventually pays for itself. As AI technologies become more advanced, accounting departments will have the potential to reduce human time on certain manual or repetitive tasks, while creating bandwidth to process more work.
Change is a gradual process, and the future of accounting and AI is very much still developing. For more information, please contact us.
Published on September 04, 2019