AI changes how accountants work, not what they do

When the AICPA National Tax Conference first spotlighted generative AI in November 2023, tax professionals were still trying to understand how the powerful technology might reshape tax practice.

“My hope is that AI simply makes us smarter, not replaceable,” David Lifson, CPA, Senior Advisor at Crowe, said during that year’s keynote presentation, “Are You Future Ready? ChatGPT, Generative AI and the World of Tax.”

Three years later, he’s more convinced than ever that AI cannot replace humans.

“The more I see about AI, the more I realize that it is just an evolutionary tool, that it does not have judgment,” Lifson said in an interview with the JofA ahead of the 2026 conference. “It’s a data accelerator, and I stand by everything I said three years ago.”

OpenAI introduced ChatGPT about a month after the 2022 AICPA National Tax Conference. Since then, AI has steadily gained visibility at the conference. Two sessions at the 2024 event included AI in their titles. The 2025 conference featured an AI keynote and another AI-focused session. In 2026, AI is listed among the conference’s six hot topics and appears in the titles of four sessions.

A bigger shift

At this year’s conference, AI is the focus of sessions on ethics, workflow, firm operations, and the changing role of practitioners. As the technology becomes a larger part of practice, tax professionals are assessing it against earlier innovations that reshaped the profession, including spreadsheets, e-filing, and online tax research.

One conference session this year is “What in the AI Is Happening With Tax Firms,” led by Brian Davis, CPA, CEO of One Stop CPA and founder of OneForm for Accountants, and two other CPAs. The session examines how firms are adopting AI tools and where practitioners are finding value.

Davis approaches the topic from two points of view. While he sees AI as bigger than previous advances such as cloud computing or e-filing because it is “in every area of the business,” he also thinks practitioners routinely overestimate what the technology can do.

“It’s sometimes definitely overhyped,” he said. “It’s not the flip-on switch that I think everybody kind of makes it seem to be.”

Changing the work

Both Lifson, who at age 76 has spent five decades in the profession, and Davis, 38, view AI not as a break from the past but as the latest step in a progression of technologies that have changed how accountants work. Spreadsheets replaced columnar pads. Electronic research replaced shelves of tax books.

Technology historically has not reduced the amount of work accountants perform so much as changed the nature of the work, Lifson said. As firms became more efficient, client expectations rose, the tax code grew more complex, and the number of choices requiring professional judgment increased.

“Every time you became more efficient, the world became more demanding,” he said.

Davis has seen that change play out in his firm. In the past, he would say he had 250 clients when that meant he handled 250 tax returns.

“Now I count the relationships, and you can have less than 50 relationships and do more than those 250 tax returns that I used to call clients. When you see the economics behind it, it’s eye-opening,” he said.

While AI isn’t wholly responsible for the change, which also improved his work-life balance, he said it did accelerate his planning, turning what had been a five-year road map into one accomplished in a single tax season.

Experience (still) matters

Even with those gains, Davis said practitioners should be realistic about what AI can and cannot do. AI can be transformative, but it does not solve every problem.

And that’s a good thing.

“If there were some technology that could just replace everything that I’ve done and all my accounting skills, then I’m doomed,” Davis said. “It’s powerful. But there’s definitely still a difference between somebody with two years of experience and somebody with 20.”

The shift is affecting how firms train staff, conduct research, communicate with clients, and build advisory practices. Work that once took years to master can be learned faster with the help of AI tools, he said.

Safeguards and risks

Looking back, Lifson said one issue the 2023 conference panel did not fully appreciate was the need for safeguards as AI systems became more powerful and autonomous.

The advice that firms ensure humans maintain a key role in AI use has evolved into a broader discussion about how AI is integrated into workflows, how professionals use it to improve their work, and what guardrails are needed to limit mistakes, he said.

“If you use AI to do the first stage of any work, you have to make sure that the person reviewing what the AI did is trained, informed, and skillful enough to find AI’s mistakes,” he said.

Davis expressed a similar concern, warning that overreliance on AI can create “false confidence.”

“It’s one of those things where nobody knows,” he said. “We don’t know where this is going. We don’t know. Is it going to wipe out an entire industry? Is it going to make the world a better place? We’re all kind of going through that together.”

Lifson worries about the global risk of AI for banks, governments, or other entities “because we become so reliant on computers to process information and transactions and store the data. Paper books and records are disappearing.”

“But that’s a different discussion to me. The risk of dangerous AI is different than the use of AI in human life,” he said. “This makes us better. That’s my takeaway. It makes us better.”

— To comment on this article or to suggest an idea for another article, contact Martha Waggoner at Martha.Waggoner@aicpa-cima.com.