Most people picture AI as a chatbot that answers your questions. In accounting, however, some of the most important changes are happening quietly inside the software. Modern systems can read receipts, extract invoice information, suggest account categories, match bank transactions, flag duplicates, draft reports and identify unusual activity. Tax software can import documents, carry information to the correct schedules and guide taxpayers through relatively straightforward returns.

None of these features are perfect, and we should review each one of them. Together, however, they can allow one person to complete work that once required several hours, or several employees. When software reduces the time required to produce the same output, clients eventually question why they should continue paying the same price for a reduced effort.

This is where the threat becomes real. AI does not need to replace every part of a job to change the economics of that job. It only needs to automate enough of the process so fewer people are needed, and to make clients unwilling to pay the same rates they paid for the processing in the past.

Routine bookkeeping is particularly exposed because much of it is repetitive and rules-based. Categorizing transactions, matching deposits, entering bills and generating standard reports are exactly the kinds of tasks software developers want to automate. Current AI tools can automate up to 80% of routine bookkeeping and transactional tasks.

The U.S. Bureau of Labor Statistics projects employment of bookkeeping, accounting and auditing clerks to decline 6% from 2024 through 2034. It still expects about 170,000 openings a year, mostly because workers will retire or move into other occupations. That is not the disappearance of bookkeeping. It's a warning that the work is changing.

A business will still need accurate books of course. The difference is that the professional may spend less time typing information and more time reviewing exceptions, correcting software mistakes, reconciling complex accounts and explaining what the numbers mean.

The same Bureau of Labor Statistics projections tell a different story for accountants and EA’s. Employment is expected to grow 5% from 2024 through 2034, faster than the average for both of these occupations.

Why would bookkeeping decline while accounting grow? One explanation is that technology is reducing lower-level transaction processing while increasing the need for analysis, compliance, controls and judgment. Currently, AI can draft an explanation of a cash-flow problem, but it does not know whether the owner is planning an expansion, hiding a weak month or misunderstanding the report. Thinking like a human still matters.

Of course, that does not mean accountants can ignore AI. A professional who takes five hours to perform work that an AI-assisted competitor can complete and review in two hours is going to struggle to compete. The threat may come less from AI itself and more from another professional that knows how to use AI tools effectively.

Tax preparation has two very different sides. On one side are uncomplicated returns involving W-2 wages, interest and common deductions. These returns are becoming easier for taxpayers to prepare through guided software. AI will make the preparation even more repetitious.

On the other side are returns involving businesses, rental property, basis calculations, multiple states, payroll-tax problems, notices, collections and uncertain facts. These matters require more than just filling out forms. They require asking questions, documentation, research and professional judgment.

Enrolled Agents and CPA preparers may therefore see pressure at the simple-return end of the market while demand remains for planning, controversy and complex compliance. A return preparer also remains responsible for the work. AI cannot sign the return, interview the taxpayer or stand behind a position during an IRS audit.

The IRS itself is integrating advanced analytics and emerging technologies into tax administration. That could help the agency detect inconsistencies and select issues for review more efficiently. Tax professionals may eventually be using AI to prepare and review returns while the IRS uses its own systems to analyze them. Accuracy and documentation will become more important, not less.

The first mistake is assuming that an AI-generated answer is correct. Tax law is filled with exceptions, effective dates, thresholds and facts that often change. An invented citation or outdated rule can create a costly penalty for a client. AI output should be treated as a draft or research lead, never as final authority.

The second mistake is pasting confidential client information into a public or unapproved AI system. Tax records may contain Social Security numbers, bank information, payroll data and other sensitive information. Public AI tools store the information you type or paste into them, but they rarely make it directly public. 

Instead, companies often store your chats on private servers and may use your inputs to train future models, meaning private details could accidentally leak into future AI responses or face security risks. The IRS publishes safeguarding guidance for tax professionals, and using AI does not remove those obligations. Firms should know what a tool stores, how the information is used and whether it is approved for confidential data before entering anything identifiable.

One practical step is to separate your services into two parts, processing and judgment. Processing includes data entry, routine categorization and standard report preparation. Judgment includes cleanup decisions, tax planning, account design, internal controls, cash-flow interpretation and IRS representation. If nearly all your revenue comes from the first part, you need to begin developing at least one service from the second part.

A second step is to create a human-review checklist for every AI-assisted workflow. For example:

These steps make technology safer and help the professional remain responsible for the final result.

There is an uncomfortable truth here. Some tasks that professionals have billed for may no longer justify the same number of hours. There is also an opportunity. If automation removes several hours of transaction processing, those hours can be used to identify a margin problem, prepare a tax projection, clean up the chart of accounts or explain why the business has profit on paper but little cash in the bank.

Clients rarely wake up excited to spend money on reconciliation services. They want to know whether the business is profitable, whether taxes are under control and whether they can afford the next purchase. AI may reduce the cost of creating a report while increasing the cost of someone who can interpret them.

AI will probably be bad for professionals who continue selling only work that AI can perform faster and more cheaply. It can be good for professionals who use it to improve accuracy, responsiveness and insights while keeping a qualified person in control.

The profession is not disappearing. It is moving up the value chain. Bookkeepers can become reviewers and systems specialists. Accountants can spend more time analyzing and advising. Tax preparers can develop planning, representation and specialty expertise.

The good thing is you don’t have to become a programmer to remain valuable. Start by learning at least one useful tool. Protecting client information and reviewing every result carefully. The goal is not to compete with AI. The goal is to use AI while providing the judgment and accountability that clients need.

If you are unsure whether your accounting process is giving you useful information, or whether automation is creating errors no one has reviewed, The Smith Advisory can help you take a closer look. Feel free to contact us at [email protected] to discuss how stronger books, better review procedures and practical tax guidance may help you make decisions with more confidence.

This article is for general informational and educational purposes only. It is not intended to provide accounting, tax, legal, investment or technology-security advice for any specific person or business. Tax rules and technology change, and individual circumstances vary. Consult a qualified professional before acting on information discussed in this article.