Study finds accountants slow to adopt gen AI

When asked to rate how much they use generative AI tools on a scale from 1-10, accountants rated their overall use at 1.56, indicating a very low adoption of the new technology. 

This is according to a recent paper that appeared in Accounting Horizons, authored by University of Miami academics Mark Ross and James Zhang. The data was drawn from 136 practitioners who completed a survey about generative AI usage. It found that not only did accounting professionals rate their overall usage of generative AI at 1.56, they rated its current use in their jobs at only 0.990. This was despite poll respondents generally agreeing that their organizations were at least somewhat encouraging of their use of these tools for appropriate tasks, rating this at 4.390. 

This tracks with another finding in the data: compared with lower level employees, leaders displayed a much stronger intention to use generative AI, and they perceive the technology as overall more useful. They also find it easier to use. 

The data also found that those in public accounting firms are more likely to plan to use generative AI in the future but, at the same time, staff don't seem very enthusiastic about it, as their attitudes on the technology is lower than it is for those in other organizations. The researchers theorized that this is due to the highly regulated nature of public accounting, particularly around the vast amounts of confidential client information they maintain. 

Of those who are using generative AI, it was primarily to draft memos, reports and emails as well as to research technical issues like Treasury regulations or GAAP standards and identify best practices for administrative or business tasks. 

The study's results stand in contrast to data from other, similar surveys. While the survey took place last summer, other surveys from the same time showed a higher use rate. Around August of last year, Earmark CPE founder Blake Oliver at the time noted that 50% of executives stated they are currently using AI in their finance and tax departments already. Meanwhile, in December, another survey found that 83% of corporate tax and auditing professionals were aware of AI use in their own workplace. A Karbon study that polled accounts around November also found that those not using AI at all numbered just 22%. 

However, wild swings in data from one study to the next does not seem very unusual when it comes to AI. For instance, while this QuickBooks survey says 98% of accountants are using AI, and a KPMG one shows 72% are working it into their financial reporting process, a Rightworks poll found 73% of accountants are not using AI and 35% have no plans to.

In an email, Zhang, one of the study's authors, theorized that differences between their data and others might come down to their sample, which came from members of state CPA societies, and so overrepresents public accountants in professional organizations; in contrast, the term 'accountant' is very broad and includes many different kinds of professionals, and so "different studies use different samples to proximate accountants but none is unbiased, resulting in differences in conclusions." 

He also noted that his own paper was subjected to three rounds of intense double-blind peer review, as it was published in an ABDC A-ranked journal. Other less formal studies may not have gone through the same review process as them. He said he believes the study is the first professional survey study on accountants' usage of Gen AI. 

There is also the issue that, sometimes, what exactly a survey means by AI can be vague. While generative AI is a specific thing, accountants have been using AI systems in general for a long time. To make sure the conclusions are clear, he said, the study has to define what it really means by AI. 

For the authors' own part, they used "ChatGPT" throughout the paper just as an example to represent Gen AI tools overall, given its popularity.