With businesses busying themselves for the close of tax season, many may be taking second glances at how emerging technologies like artificial intelligence can help them speed through complex and important financial reports more efficiently.
However, for tax and finance professionals to embrace AI, they have to trust what the technology tells them: AI-supported programs must not only be able to understand tax forms, but be able to quickly and easily explain the conclusions they draw, said Ahmad Ibrahim, CEO and co-founder of automation and tax software provider Neo.Tax.
“I think ultimately the success of your AI product is going to come down to…obviously people need to trust it, right?” Ibrahim said in an interview.
Founded in 2019, the Mountain View, California-based company provides automated tax programs that enable companies to quickly file for the R&D tax credit, according to its website. This is a federal tax credit available for small businesses with less than $5 million in revenue for the tax year among other requirements, according to the Internal Revenue Service.
Solving the AI user problem
Acknowledging what AI can and cannot tell you is essential to generating user trust, said Ibrahim, an alum of global finance service provider Intuit where he served as a product manager for two years, according to his LinkedIn profile. This could be as simple as being upfront with the user when you lack high confidence in what the tool says, he added.
Continued experimentation with AI by big name companies has put a spotlight on the technology’s potential, but it is essential to pay close attention to the technology’s shortcomings, especially when it comes to heavily scrutinized spaces like financials — “you can’t hallucinate in taxes,” Ibrahim said.
OpenAI warned upon launching its latest version of its tool that ChatGPT 4.0 was still prone to “hallucinating”— a term that refers to the tool creating incorrect facts — and still was not fully reliable, urging “great care” when deploying the tool especially in “high-stakes contexts.” Demos of other AI solutions, including both Microsoft and Google’s tools, also displayed serious errors, including providing inaccurate financial figures when asked to provide them from earnings reports, according to a February report by Fortune.
Building trust therefore represents a “design problem of how quickly you can surface” an explanation about how the software arrived at its conclusions, Ibrahim said.
For it to be embraced in the finance sector, it’s also important to have an understanding of where AI can provide the most benefit. Pointing an AI tool at the tax code is less useful than one might think; where the AI should be pointed instead is at the business itself, Ibrahim said, where it can automatically put transactions and details into context of that code.
“If you can point it at all those things, it's almost like an accountant reviewed every single document in the company,” he said.
Though the introduction of tools like ChatGPT has also made the AI space much more competitive, “we feel fortunate that we've already been feeling around in some of the obvious places where you can use AI and those don't bear as much fruit as you think,” Ibrahim said. “And so we have the added benefit of knowing where to point it.”
AI and the accounting future
Finance professionals are already convinced of AI’s impact: a recent survey found 64% of CFOs and VPs of finance believe AI and automation will have a bigger impact over the next 10 years than other technological developments. Yet many firms continue to cling to their traditional tools, with about 93% of respondents still using Excel on a daily basis, the study found.
The lack of trust when it comes to AI is understandable when it comes to company finances; at the end of the day, the company is still the one on the hook for any errors, not the tool, Ibrahim said. However, some accountants may be pushing back because of a fear that AI and automation are going take over such roles, Ibrahim said.
With less and less individuals coming into industry or taking the certified public accounting exam and the current accounting population aging, the industry has a responsibility to fill the widening gap, making it likely that looking into the future, “much of accounting does get automated away,” he said.
“That apprehension and worry I think has always been there and I think it's justified,” Ibrahim said. “I think the best accountants will figure out ways to up level and employ AI to their advantage.”
It’s important to find “accountants that are living in the future,” he said. Many are already expecting this melding of accounting and IT skills as shortages continue, CFO Dive previously reported. Accounting is just one area where AI might have an effect over the future of the finance function.
It could also impact the role of the CFO, though Ibrahim does not see that job going away. Rather, it will put more emphasis on the creative aspects of the job as opposed to the strictly quantitative, he said. More value could be placed on CFOs who have banked decades of experience, allowing them to draw conclusions AI cannot.
One could think of AI like a hyper-intelligent college student, Ibrahim said, who has read and can regurgitate all of its finance textbooks, but “it's never seen what happens when you try to present this (data) to the street,” Ibrahim said.