- news
- TECHNOLOGY
CPAs at a crossroads: Real approaches to artificial intelligence
Related
Summing up economic sentiment and concerns about inflation and tariffs
Incorporating prompt engineering into the accounting curriculum
Create a dynamic to-do list with Excel’s checkboxes
Editor’s note: Those who purchased an all-access pass to the Digital CPA Conference 2023 can now view this and other archived sessions. If you didn’t attend DCPA 2023, you still can access this session.
If a former firm partner turned artificial intelligence (AI) expert feels overwhelmed with the current rate of change related to generative AI (Gen AI) tools like ChatGPT, where does that leave CPAs charged with advising clients, work teams, or both?
“The rate of change right now is frankly exhausting,” Jason Staats, CPA, said during his session last month at the Digital CPA Conference. “It’s really hard to keep up with, and it may require a different level of investment and learning than we’ve had to make before.”
When it comes to ChatGPT and similar content-creating AI tools, Staats believes CPAs who advise others on a day-to-day basis are fast approaching a crossroads.
“I feel like anytime there’s like a big hard new thing, especially if you work with small businesses, it seems to fall on the accountants,” he said. “Simultaneously, that’s the best and the worst part of doing what we do. But it’s worth zooming out to think about how much bigger this is than just us.”
During his DCPA session titled “Empowering Teams With ChatGPT,” Staats shared why CPAs need to get a handle on emerging technologies and how they can start doing so today.
AI is too big to ignore
“If this truly is this civilization-shifting thing, that’s going to be a pretty darn big blind spot when it comes to how we work with our clients if we’re not plugged into it,” Staats said. “Keep in mind that an investment in yourself is an investment in your clients.”
AI is supposed to make things easier, not harder, but designing an adoption plan for work teams or answering clients’ questions are challenges in the short term to be sure.
The good news? You don’t have to go it alone.
AI is too big for one person
Staats recently helped CPA.com, the AICPA’s business and technology subsidiary, design a Gen AI toolkit that serves as a good starting point. He said that ChatGPT (and similar tools he discussed at DCPA) is “a tool for all of us” that, with good planning, can empower users up and down the company ladder to help implement.
“The one thing you absolutely need to be doing is optimizing for agility. To me, that means enabling bottom-up adoption,” Staats said. “That means giving people the tools they need to start experimenting, and rewarding curiosity in a way that’s visible.”
Encouraging your teams to experiment with ChatGPT and other technologies is part of the equation, but Staats said it’s just as important to encourage them to share their discoveries along the way.
“Those little nuggets that you stumble into and you’re like, ‘That’s going to save me like an hour every month’ — the more we can share that stuff, the more that you don’t just have to figure it all out yourself,” Staats said. “How can we eliminate barriers to share more of that stuff that we’re learning along the way? It can be as simple as spinning up a group text or a Slack channel or a Teams channel or something like that.”
AI requires guardrails (but wide ones)
“The policy doc, this is honestly the Holy Grail. If there’s one thing that you take out of this, set up a policy doc,” said Staats, who estimated that 80% of firms don’t yet have one.
While the general thought is that policy documents tell users what they can’t do, Staats advocates for a document designed to emphasize what they can do. While the creation of sound policies may necessarily involve your IT team helping define what data within different categories (sensitive, sensitive-redacted, and nonsensitive) shouldn’t be shared with ChatGPT, it’s also about what data should be shared.
“It should define what’s in bounds, so people can go crazy,” Staats said. “You’ve got to say, ‘This is the OK way to use language models with this type of information. Now go wild.'”
This is only the beginning
“This is the easiest it will ever be,” Staats said of the ever-evolving AI space. “I know we’re feeling overwhelmed, and this is hard and the rate of change continues to increase. But this is just where we’re headed. As a result, agility becomes super important.”
Staats’ simple advice in the face of the complexity?
“A lot is going to change, but it’s just going to mean plugging into what’s changing. However you learn best — whether that’s podcasts, social media, CPE — keep engaging in that stuff.
“And just focus on, ‘How can I be useful this week for my clients and the team that I support?'”
— To comment on this article or to suggest an idea for another article, contact Bryan Strickland at Bryan.Strickland@aicpa-cima.com.