The accounting world has a complicated relationship with new technology. We adopted Excel religiously, then spent the next thirty years being suspicious of everything else.
AI is no different. All accountants I talk to fall into one of two camps: they either dismiss it as a gimmick that can’t handle the nuance of real financial work, or they’re convinced it’s the final push that makes them replaceable, by AI, by offshore teams, by whoever’s cheaper and faster. Both reactions miss the point.
The mistake isn’t skepticism. Skepticism is healthy in this profession. The mistake is thinking AI is a product you evaluate once and either adopt or reject. It’s not. It’s a skill that you build. And the window to build it quietly is closing.
Here’s what I’ve noticed after spending the last few years building AI-driven workflows for small and mid-size businesses: accountants are actually better positioned than almost anyone to establish this skill.
The profession already requires breaking complex problems into discrete steps, validating outputs against known rules, and knowing when a calculation doesn’t look right. That’s exactly the skillset you need to use AI without getting burned by it.
Here’s the uncomfortable part. The profession built on verifying everything is treating AI as something you either trust completely or avoid entirely.
That said, some are getting it right. The ones extracting real value treat it like a competent junior staff member: useful, fast, occasionally overconfident, and always in need of a review.
That one shift changes the whole question. The mechanical tasks were never the soul of the work. They were the barrier to it.
When AI provides the speed, the CPA provides the trust.
The CPA license isn’t just a credential. It’s a standard of verification. That was true before AI. It’s more true now.
The judgment was always the point.
The spreadsheet isn’t going anywhere. But the person who only knows how to fill it in probably is.