When it comes to artificial intelligence in accounting, the future is already here but not everyone seems to have noticed.
Speaking at the Information Technology Alliance's spring collaborative in Memphis, Tennessee, Gregg Landers — managing director of client accounting and advisory services and internal control services with Top 10 Firm CBIZ — pointed out that while much of the conversation around AI focuses on what it might do in the future, many of those capabilities are already here. What's missing is coordination.
"I think the technology to actually do all these things exists today, they're just not all in one place packaged together. All the components of AI–generative, machine learning, agentic, all those pieces individually are here, now," Landers said.
He rattled off a list of things that AI can already do today. There is already technology to take any PDF, break it down, look at the data and estimate where it should be posted in the ledger; or to examine and analyze financial data whether for fraud risks or financial opportunities; or to make suggestions or ask deep questions to guide future business objectives; and much more.
"All this technology exists today, in separate little pieces all around the place. And it frustrates me that our different vendor friends can't put it in one package and eliminate the waste here so we can automate a lot of these core accounting elements. … It frustrates me we're still having this conversation," he said.
While such tools are anticipated to be developed, he noted that people have been promising accountants such solutions for ages, and while salespeople might say the capacities Landers are seeking are just over the horizon, twenty years later he is still having the same conversations with vendors.
Chris Gallo, director of outsourced business accounting services with Kansas-based firm Creative Planning and another one of the panelists, suggested economic incentives keep companies from developing the kind of single universal AI solution Landers wanted. If there were single universal AI systems that combined every possible capability available today, and if such a system were easy to implement, then it would be very easy to convert and migrate data, which would make it very easy to switch between vendors and therefore threaten their business models. The rise of a truly holistic universal AI system, he speculated, isn't a matter of technology but economics.
"If AI will do what [boosters] say it will do, what prevents any business owner from pushing a button and going from Sage to NetSuite to Oracle? And the implementation will just be a push of a button, [so] what keeps me from changing every month, changing my support codes?" Gallo asked. "They already know how to do a lot of that stuff but monetizing it would be an impossibility, they would die, so we're never going to get a single holistic program no matter what tech wants to tell you."
But regardless of what is down the line, Gallo echoed Landers in pointing out that while so many are talking about how they could use AI in the future, they don't seem to be aware of how much they, themselves, have already been using AI for years, even before the generative-AI explosion in late 2022. The data analytics programs accountants use to track data like receivables and extrapolate from trends use AI, as does the software that automates timecard entries at the end of the day and the tool that color codes risk factors on a spreadsheet.
"I would say all our tools use AI in some fashion. Everything you use probably has some sort of AI component in some sense," Gallo said, adding that it may not fit what certain people nowadays think of as AI. But even if it's not, it's still extremely useful for accountants, more so than just querying ChatGPT. "There's just such a vast difference in asking a large language learning model, LLM, questions and finding answers, getting data out in the AI world, and actually using the tools that we've got right now in front of us."
Although the last panelist, Jessica Barnas, the partner leading the finance and accounting solutions team advisory group for Top 25 Firm Wipfli, noted that even these generic LLMs can be useful for CAS practitioners. Noting the profession's well-known capacity problems, she said she doesn't mind if clients refer to things like ChatGPT for basic information.
"When we talk about talent and having limited capacity, I wish clients would attempt to ChatGPT the answer themselves before coming to us with these lower level questions. For years we've been trying to engineer solutions like chatbots [to be like] a help desk system, with a ticket-run resolution and if the questions become more complex it gets escalated to a human. … I think you're probably going into a direction of being scared that it would happen but I would gladly embrace that happening," Barnas said.
She added that people should still exercise skepticism about the information they get from AI, though this applies to any source; people likely are coming up with silly ideas about, say, what can and can't be deducted based on ChatGPT conversations, but at the same time she said they also get silly ideas from TikTok and Instagram, too. Overall, it is important to educate not just staff but clients as well on being skeptical of what they hear.
