Jatin Narang, CEO of Verito.com, a leader in private hosting & managed IT solutions, empowering accounting firms in the digital era.

getty

Every white-collar profession is having a similar conversation right now. AI is eating the entry-level work that used to train the next generation. Consulting firms are writing about it in Harvard Business Review. Law firms are debating it in Above the Law. Investment banks are building systems like Project Mercury to automate the first two years of an analyst's job.

Accounting's version of this problem is not like the others. It's much worse. And the reason is demographic, not just technological. Accounting is being compressed from both sides, while AI eats away at the ladder.

The Collision Being Faced

When we look at four key professional services industries having this conversation right now, a stark divergence emerges.

In law, there's an oversupply of associates. More law students graduate each year than Big Law can absorb. Similarly, consulting has hundreds of thousands of applications a year. McKinsey, Bain and BCG could redesign the analyst role 10 times over and still fill seats. In investment banking, OpenAI's Project Mercury automates junior analyst work, and the replacement applicants are already interviewing. ​

Accounting is different. The AICPA reports that 75% of current CPAs are eligible to retire within 15 years. Over 300,000 accountants have left the profession since 2019. CPA exam candidates fell from roughly 50,000 in 2010 to 32,000 in 2021. The pipeline has been shrinking for a decade, before AI became a factor. ​

Now add the AI pressure. Stanford Digital Economy Lab research published in 2025 found that early-career workers (ages 22–25) in AI-exposed occupations, accounting included, experienced a 16% relative decline in employment since the widespread adoption of generative AI in late 2022.​​

Accounting does not have the benefits of other industries. Seniors are leaving faster than training can keep up. Meanwhile, the pool of replacements has been shrinking for 15 years. AI isn't creating accounting's succession crisis. It's arriving in the middle of one.

What We're Hearing From Firm Owners

Working with more than 1,000 tax and accounting firms at Verito, the concern I hear most isn't about cybersecurity or software costs. It's about succession.

The questions and concerns arrive in different forms. Who reviews the returns when I retire? My senior manager is competent but she's never actually prepared a complex trust return. The juniors we hired over the last two years are faster, but they can't tell when the software is wrong.

That last one is the specific thing Stanford and MIT measured. It's also the thing firm owners are noticing in their own offices without knowing there's research on it.

What Junior Work Actually Builds

Senior accountants treat AI as a collaborator. They push back when the system's confidence drops. They catch errors and apply judgment. Junior accountants are more likely to accept AI output at face value, even when the system flags its own uncertainty.

That finding from Jung Ho Choi at Stanford GSB and Chloe Xie at MIT Sloan isn't a generational critique. It's a reps problem. The seniors know what a real answer looks like because they've produced thousands of returns. The juniors haven't yet.

Project that forward. Today's juniors become tomorrow's seniors without developing the judgment their predecessors did. Every K-1 a senior miscategorized at age 25 became a pattern they'd catch at age 45. Strip that work out and you don't just lose headcount. You lose the mechanism that makes reviewers competent to review.

The Signature Problem

Consulting partners don't personally sign deliverables that carry federal penalties. Investment bankers don't either. Law firm partners sign pleadings, but liability is distributed across the firm, unlike for a PTIN holder.

When a CPA signs a return, the signature is theirs. The IRS can revoke her PTIN. A state board can pull its license. Malpractice suits follow the human, not the machine that drafted the underlying work. This liability structure is fundamentally different from the adjacent professions facing AI disruption.

Supervision: Not The Same As Judgment

Today's accounting situation is a structural redefinition of what junior work involves, overlapping with a retirement wave that has no modern precedent.

The response from Big 4 accounting leadership has been to redesign. PwC has told juniors they'll be expected to function like managers within three years. It assumes that the new junior role of supervising AI output produces the same judgment as the old junior role of producing work from scratch.

The Choi and Xie research suggests it doesn't. The juniors supervising AI output aren't catching the AI's errors the way the seniors who learned the old way are.

The Question The Profession Hasn't Answered

The AICPA launched its Profession Ready Initiative in February 2026 to define skills early-career CPAs need in an AI-driven market. Every major firm has an AI training track. Universities are redesigning curricula. None of these responses tackle the main issue: Seniors skilled in reviews are leaving too quickly.

The math is specific. Three-quarters of today's reviewers are eligible to exit within 15 years. The pipeline feeding replacements has been shrinking. AI is now absorbing the work that historically turned replacements into reviewers. These three curves are not independent; they're compounding.

In 2030, someone might sit down at a desk, review an AI-prepared return and sign it. The liability will follow the signature, as it always has. The judgment that used to sit behind the signature will be increasingly scarce.

The firms that navigate this well won't be the ones that adopt AI the fastest. They'll be the ones that are most deliberate about where they deploy it. Not every task that can be automated should be, at least not for every staff member. Complex return preparation, multi-entity reconciliation, trust and estate work are the engagements where junior accountants develop the pattern recognition that eventually makes them competent reviewers. Firms should identify which categories of work build judgment and protect those as hands-on assignments, even when AI could handle them faster.

The math demands urgency. Firms with partners approaching retirement need active knowledge transfer plans, not assumptions that the next generation will figure it out. That means structured mentorship, co-review processes where a senior walks a junior through the reasoning behind each decision and honest internal assessments of where the judgment gaps already exist.

I believe the profession has perhaps a decade to solve this, and the firms that start now will be the ones still signing returns with confidence beyond 2035.​​

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?