HR Block’s use of IBM’s Watson technology this past tax season is only one sign of the increasing involvement of artificial intelligence in tax preparation, with even Big Four firms like PricewaterhouseCoopers making major investments in AI for servicing their tax clients.
PwC recently released a report on the use of AI and machine learning in tax analytics. According to the report, AI tools will effectively become digital “assistants” that could potentially replace the basic capabilities of first- and second-year tax associates. Machine learning is already being used to identify historical patterns of information to automatically suggest certain tax adjustments to tax professionals.
“We’re confident that we will be seeing a lot more of AI and the components of that technology applying to tax,” said PwC principal and tax technology and process leader Michael Shehab. “When people think about artificial intelligence, they think about the movies and sci-fi. We think about it more in terms of the concept of machine learning and natural language processing. We’ve seen a lot of applications already for those tools. We think this is not a fad and has the possibility to change our industry.”
The PwC report describes several different ways that AI technology can be used by tax professionals, such as generating K-1 schedules, categorizing and processing information that have different formats but use similar forms across tax jurisdictions, and classifying images for fixed asset recognition, or meals and entertainment expenses, based on tax rules, vendors, amount, location and time of day. Other functions can include statistical modeling and answering tax research questions. A big advantage of AI is the ability to deal with vast amounts of data and use it to learn.
“In our tax practice we do a lot of value-added consulting and compliance,” said Shehab. “Associated with tax reporting are large volumes and repetitive tasks. Machine learning is a component of AI, but the low-hanging fruit is the automation of those repetitive tasks. Historically we would try to automate them through technology tools, and the human professional would still have to manage and interact completely with that technology tool. In the future, with more sophisticated tools than the ones we have deployed already that have machine learning, the technology tool has the ability to learn that repetitive task and grow over time. It’s still curated by a tax professional, but we believe most tax preparation tasks can be done by a technology tool, allowing the human to elevate themselves and to review, rather than the human being bogged down with data manipulation and manual reconciliation.”
He offers the example of book tax adjustments. “We have thousands and thousands of clients and decades of years of experience doing that,” said Shehab. “Being able to put that into a technology tool, all that historical knowledge and all that knowledge from a lot of different clients, and the historical detail from an individual client, allows the technology tool to make an arguably more accurate decision on what the book tax adjustment should be, and at least take a stab at preparing all the book tax adjustments for the tax professional. The tax professional can then begin with review rather than the baseline preparation.”
Shehab sees applications for both individual and corporate taxpayers, but especially in the corporate area. “We think it has more applicability in the corporate space because the corporate fact patterns are more complex and have large volumes, so the more complex and the larger the volume, the more applicable the typical artificial intelligence tool will be,” he said.
While there might be less need for tax preparers who are used to doing the grunt work, Shehab believes it will be beneficial to most tax professionals. “We think it’s a big opportunity for growth in our industry,” he said. “The more we can free our people up to work with our clients and our industry fact patterns, the more value we’re going to bring the clients. We’re thirsting to free them up from the data manipulation into higher-value subject matter expertise, and we don’t think an AI engine is going to replace that higher-value service anytime soon.”
Small firms may initially have trouble competing with the Big Four firms that can invest significant amounts of money in AI technology, but Shehab believes the technology will eventually filter down into commercial tax prep software for professionals.
“I don’t want to say it’s going to be exclusive to the Big Four, but I think it’s either going to be a third party large-scale technology vendor, or someone in the Big Four, or a superregional firm that will play in this area,” he said. “I think it will be tough for a 100-person accounting firm to make this type of investment, but once a couple of people make the investment, they’re going to seek to monetize it, either through licensing it to their peers or other folks, or a third-party software firm might come in and try to drive the charge. But that is the unique thing about AI. It requires a lot of investment and a lot of patience because it’s a multiyear journey.”
The technology is already making an impact at PwC. “I know at PwC it’s going to be an absolute reality that we will have elements of AI, whether that be machine learning, natural language processing or more predictive analytics across all aspects of our business,” said Shehab. “It’s going to reshape how we spend our hours, but it’s not necessarily going to replace professionals. We view AI as a digital assistant.”
AI is already being used in PwC’s proprietary internal tax preparation systems, as well as analytics. “What has been missing in this industry for a long time, and what we’re really focused on, is not just delivering a tax return but delivering the analytics associated with the tax return,” said Shehab. “We’re trying to make the tax return preparation process more efficient, but we’re also trying to make it higher value added rather than simply delivering a tax return.”