340,000 Accountants Left. The Profession Is Still Growing. Here Is Why.

The Bureau of Labor Statistics projects 5 percent growth in accountant and auditor employment over the 2024 to 2034 decade, faster than the average for all occupations. Over the same period, it projects a 6 percent decline in bookkeeping, accounting, and auditing clerk positions. Goldman Sachs, in a separate analysis, rated accountants and auditors at the highest risk of AI displacement based on task exposure. All three findings are accurate, and they point in the same direction once you separate what the data is actually measuring. This article works through 21 data points from government statistics, academic field studies, and professional body surveys to explain why the profession is contracting in some places and growing in others. For more background statistics, the 75-stat roundup covers the broader AI-in-accounting picture.

Are accountants being replaced by AI?

The Goldman Sachs finding and the BLS projection look contradictory until you see they are measuring different things. Goldman Sachs analysed task exposure: what percentage of an accountant's daily tasks could theoretically be performed by AI given current technology. The BLS projection measures expected employment headcount across the decade. Task exposure tells you which tasks a role contains that AI could potentially handle. Employment projection tells you whether firms will employ more or fewer people in that role overall. A role can have high task exposure and growing employment at the same time, if the value of the remaining human work increases as the automatable tasks are handled by software.

The Bureau of Labor Statistics projects 5 percent growth in accountant and auditor employment between 2024 and 2034, faster than the average for all occupations. (Bureau of Labor Statistics Occupational Outlook Handbook)

Over the same decade, the BLS projects a 6 percent decline in bookkeeping, accounting, and auditing clerk positions. (Bureau of Labor Statistics Occupational Outlook Handbook)

Goldman Sachs rates accountants and auditors as facing the highest risk of displacement by AI based on task exposure analysis, not employment projections. The distinction matters: task exposure measures theoretical automation potential, not actual job loss. (Goldman Sachs Global Economics Research)

Both BLS trends run along the same dividing line: tasks AI handles well today versus tasks it does not. Clerks spend most of their time on data entry, transaction coding, invoice matching, and document processing. That work is high-volume, pattern-based, and structurally well suited to machine learning. Accountants spend more of their time on interpretation and professional judgement, work that depends on client context. AI is absorbing the tasks in one role and leaving the other intact.

Why did 340,000 accountants leave?

The departure started before AI tools were production-ready for accounting work. People left because of burnout, a retirement cliff that has been building for two decades, and a pipeline that dried up as fewer students chose the profession. The people who stayed did so through a period of mounting pressure that has not fully resolved.

340,000 accountants left the US profession between 2019 and 2023, a 17 percent decline from approximately 1.6 million. The figure comes from the largest workforce study the profession has conducted: approximately 8,000 survey responses and briefings with 15,000 participants. (AICPA/NASBA/NPAG Accounting Talent Strategy Report)

Accounting graduates hit a 20-year low in 2023 to 2024: 55,152 graduates, down 6.6 percent year on year. Master's degrees in accounting fell 15 percent over the same period. (AICPA/CIMA 2025 Trends Report)

Nearly 75 percent of CPAs in the United States are at or near retirement age, according to AICPA pipeline data. The credential figure is specific to the US credentialing system. (AICPA pipeline data, reported by Ramp)

86 percent of accountants report burnout, and 25 percent say they are seriously considering leaving the profession within the next year. (Sage "Practice of Now" survey, n=1,000 across 6 markets)

None of those four numbers have anything to do with AI. Firms that lost staff to burnout in 2021 and 2022 lost them to overwork. The workforce is now smaller and older, and the same volume of compliance work still needs doing. AI tools are entering that environment, which means they are filling a gap rather than displacing people from stable jobs.

What is happening to bookkeeping and clerical roles?

Roles the BLS expects to decline are built around tasks that fit AI's current strengths: data entry, transaction categorisation, invoice matching, bank reconciliation. High-volume work on structured data with predictable rules. Roles the BLS expects to grow depend on interpretation, advisory work, and professional judgement, which AI still handles poorly. What matters is whether automatable tasks make up most of a given job or just a fraction of it.

42 percent of finance and accounting tasks can now be automated using current technology. (McKinsey Global Institute)

27 percent of all current work hours are projected to be automated by 2030, with accounting identified as facing the greatest disruption among white-collar professions. (McKinsey Global Institute)

Manual invoice processing costs $15.97 per invoice; automated processing brings that to $3.24, roughly an 80 percent reduction in per-invoice cost. (IOFM / Ardent Partners, reported by Precoro)

That invoice figure is worth sitting with. When the most time-consuming task in a job can be done for $3.24 instead of $15.97, the case for hiring a person to do only that task weakens considerably. The roles that hold up are the ones where a person handles what the software cannot: exceptions, client contact, professional review, and calls that require context the system does not have.

How is AI changing the work that remains?

The Stanford GSB and MIT Sloan field study published in 2025 by researchers Choi and Xie is the most useful data on this question. It tracked 277 accountants across 79 firms and observed actual outcomes, rather than asking respondents what they believed or intended. Field studies of this kind are harder to run than surveys, and the sample is smaller, but the findings are grounded in what happened in practice.

AI-using accountants cut 7.5 days off their monthly close. (Stanford GSB / MIT Sloan, Choi and Xie 2025, n=277 accountants + 79 firms)

The same accountants supported 55 percent more clients per week compared to peers not using AI tools. (Stanford GSB / MIT Sloan)

8.5 percent of working time shifted from data entry to advisory and review work, approximately 3.5 hours per week moved from manual processing to client-facing tasks. (Stanford GSB / MIT Sloan)

Cutting 7.5 days from the monthly close saves time. Shifting 3.5 hours a week from data entry to advisory work changes what the job actually involves. That second number is the one that reshapes the profession. The full study data is in the 75-stat roundup.

Advisory now accounts for 13 percent of firm revenue, up from 10 percent in 2024. 93 percent of firms now offer some form of advisory service. (Wolters Kluwer Future Ready Accountant Report 2025, n=2,768 across 14 countries)

Three percentage points sounds modest. But advisory was barely trackable as a revenue category in most small firms three years ago, and the Wolters Kluwer data suggests it is now a tenth of firm income across 14 countries. Compliance work takes less time when AI handles routine steps; the hours freed up have to go somewhere, and advisory is where they are going.

Who is adopting AI fastest?

The generational data here goes against the obvious prediction. Younger accountants use AI more often. Older practitioners use it for client work where their experience helps them spot when an AI output misses something. Frequency and quality of use are not the same thing.

83 percent of chartered accountants aged 18 to 24 use AI at least weekly. (Chartered Accountants Worldwide / Ipsos, n=2,718 across 48 countries)

80 percent of accountants aged 18 to 24 feel confident using AI tools, compared to 55 percent of senior decision-makers. (Chartered Accountants Worldwide / Ipsos)

One UK survey found the opposite pattern when looking specifically at client work: 50 percent of accountants aged 55 and older use AI for client insights, compared to 16 percent of those aged 18 to 24. (AccountingWEB UK) This is a single survey, smaller and less rigorous than the Chartered Accountants Worldwide data, so treat the number as a signal. But the direction makes sense. Senior practitioners have the client relationships to apply AI to, and the professional experience to know when an AI output is right, wrong, or incomplete. Junior staff use AI more often, mostly for research and drafting, not for the conversations that generate revenue.

For firms thinking about where to direct training budget, this argues against concentrating it entirely on junior staff. A senior practitioner already using AI for client work probably generates more value per hour of training than a junior employee learning general AI literacy. Specific tool training for people who already have context beats broad courses for people who do not.

What is stopping firms from adapting?

The perception gap in this data is wide. Finance professionals broadly agree that AI is consequential. Very few feel ready for it.

88 percent of finance professionals believe AI will be the most transformative technology they face, but only 8 percent feel "very well prepared" to use it. (AICPA and CIMA Future-Ready Finance Survey, n=1,446)

56 percent of finance professionals identify generative AI as the most prominent skills gap in the profession today. (AICPA and CIMA Future-Ready Finance Survey)

Only 37 percent of accounting firms invest in AI training, according to a Karbon survey of accounting professionals. (Karbon State of AI in Accounting 2025. Note: this is a vendor survey of Karbon's own user base, so it may skew toward firms already engaged with AI.)

The gap between the 88 percent who believe AI is transformative and the 37 percent investing in training is where the friction is. A firm that has seen the headline numbers and decided AI matters, but has not yet assigned time or budget to act on it, is running a risk. Tools are getting more capable. Firms that invested early will show measurably different output per person within the next two or three years. The ones watching from the sidelines will notice it when they lose staff or clients to firms that did.

Some evidence suggests the pipeline may be turning. Accounting enrolments grew 12 percent year on year for two consecutive semesters in 2024 to 2025, the first sustained reversal after years of decline. (National Student Clearinghouse, reported by Journal of Accountancy) Whether that represents a real shift in how the profession is perceived by prospective students, or a temporary fluctuation, will take several more years of data to determine. The direction is at least better than it was.

Twenty-one data points, and they point the same way: the profession is not disappearing, but its shape is changing faster than at any point in recent memory. Roles built around transactional tasks are contracting. Roles built around professional judgement and client advisory are growing. The firms that come through it are the ones that have already started closing the training gap. For the full picture across 27 primary sources, the 75-stat roundup covers market size, adoption rates, time savings, and buyer behaviour.