If an accountant uses AI to summarize hundreds of pages of documents, is that cheating?

And what if an analyst uses an AI tool to organize data or draft an initial report?

The answer depends not only on whether the tool is used, but on what we have left the tool to do for us.

AI is becoming an increasingly common part of the workplace, but its real value lies not in making employees stop thinking, but in reducing the time they spend on repetitive tasks so they can focus on work that requires human understanding, analysis, and judgment.

This principle is clearly reflected in the new guidance AICPA has issued to financial auditors on using AI in auditing.


AI can help—but it cannot take responsibility

AICPA says AI tools can improve the efficiency and effectiveness of financial reporting and audit work, but they can also introduce new risks involving data, bias, hallucinations, and the difficulty of interpreting some results.

That is why the organization emphasized one important point:

The auditor remains responsible for the audit process and the final opinion.

This idea applies to most professions. AI can suggest, summarize, search, organize, and draft—and it can even spot patterns a person might not notice quickly.

But professional responsibility cannot be handed over to it. If a model puts an incorrect figure in a financial report, an employee cannot defend the mistake by saying, “The AI wrote it.”


When is using AI helpful?

AI is especially valuable for tasks that take a lot of time but do not require a person to start every step from scratch. For example, an auditor might use data analysis tools to search thousands of entries for unusual transactions, or to help summarize long contracts and documents.

A financial analyst can use it to organize information, compare results across periods, or prepare an initial draft of an analysis that must then be reviewed. An administrative employee can use it to summarize meetings, organize tasks, or improve the wording of correspondence. In these cases, AI is not doing the job instead of the employee; it is reducing the time needed to reach the analysis and decision-making stage.


And when does it become a problem?

The problem begins when users are unable to explain what the tool has produced. If AI writes a complete financial analysis and the analyst publishes it without checking the figures or understanding the conclusions, AI is no longer just a helpful tool. It has done the real work. That comes close to professional misconduct, because the person presents the result under their own name without being able to defend it or verify that it is accurate. That is why AICPA’s guidance emphasizes the continued importance of professional judgment, professional skepticism, supervision, and review even when AI tools are used.


Speed does not mean giving up understanding

AI can be compared to a calculator. When an accountant uses a calculator, it does not mean they do not know accounting; it means they do not want to waste time doing calculations by hand. But if they do not know what needs to be calculated in the first place, the calculator will not save them. AI works the same way, but on a much larger scale. It can streamline many stages of work, but it is most valuable when used by someone who understands the field and can spot an error when one occurs.


The new skill is not avoiding AI

As these tools become more widespread, the best employee may be neither the one who refuses to use them nor the one who leaves everything to them. It is the person who knows which tasks to delegate to the tool and which decisions to keep in their own hands.

That is why the future of work does not look like a choice between people and AI. It is more likely to involve the two working together:

Machines take on more of the searching, organizing, and processing, while people remain responsible for context, judgment, ethics, and the final outcome. In this way, using AI becomes a professional skill in its own right. The question is no longer: Have you used AI?

It is:

Did you use it to help you think, or did you use it instead of thinking?