The real barrier to AI in accounting is human, not technical

Modernization will come from small and consistent steps.

Artificial intelligence hasn’t been around long, but most firms have already decided they can’t ignore it. Initial skepticism around its benefits has softened. Trust, however, is still the choke point.

Some firms jumped in too fast and paid for it.

Hallucinations cost them credibility and, in a few cases, public embarrassment. Others bought tools that promised miracles and delivered chores. The market moved faster than the profession, and AI vendors rarely sat down with accountants to explain what the technology can and can’t do. That gap killed trust before the tools even had a chance.

Deterministic AI handles repetitive, rule-based work.

It’s great at matching forms, checking numbers, and ensuring a birthdate isn’t entered as 2079. Generative AI interprets context. It can read a return and infer whether an IRA is Roth or traditional. Real accounting work needs both, plus human review. Firm leaders should explore what’s out there, but ignore anyone pitching magic.

Talent shortages and client expectations make evolution non-negotiable. Skills and systems have to change.

AI for operations

Start small. Pick one area. Test AI on tax returns your team has already completed. Compare the outputs. See what breaks. You’ll uncover discrepancies before they become liabilities.

I know, it’s a no-brainer, but many will skip this first initial step.

Use the same method for audit, bookkeeping, and advisory work. Parallel testing exposes where AI helps and where it slips. Look at differences without taking them personally.

Plenty of accountants spot their own mistakes when they compare their work to AI output. Plenty also catch the tool-making errors. Both outcomes matter. Either AI shows you a risk you didn’t know you had, or it gives you a head start you weren’t getting from older tech.

Attracting young talent

The profession’s pipeline is cracked. Most CPAs are nearing retirement, and fewer students are choosing accounting. AI won’t fix that overnight, but it can help change the job into something people actually want to do.

Bring your teams into the process. Let them list the parts of the job that drain their time and focus. Use that list to define where AI fits. Build a framework for when and how to use it, and pair it with a review process.

Run cohorts. Compare the speed and accuracy of people using AI with those of people doing the same work manually. Iterate. Retrain your models. Refine the workflow until it supports your firm instead of disrupting it.

And don’t sideline your senior staff. They carry institutional memory and networks that younger professionals don’t have yet. If junior employees complete the grunt work faster with AI, they can take on higher-level work sooner. That frees senior professionals to mentor and build relationships. Everyone moves up the value chain instead of running in place.

Who has the advantage

The Big Four are throwing money into proprietary AI. That doesn’t mean they’re ahead. Smaller firms can adopt off-the-shelf tools faster. They have less bureaucracy, less technical debt, and more room to experiment. They don’t need permission or a steering committee to modernize. That makes them the real accelerators of AI adoption in the profession.

The way forward

Modernization won’t come from one sweeping overhaul. It will come from small, consistent steps. Clear testing. Honest review. Broad participation. The firms that treat AI as a tool, not a promise, will adapt faster and serve their clients better.