Artificial intelligence has become one of the most discussed topics in the accounting profession, yet much of the conversation remains focused on a single question: Can we trust it?

The concern is understandable. Auditors are trained to be skeptical—to challenge assumptions, evaluate evidence and understand the basis for conclusions before relying upon them. However, as AI adoption continues to accelerate, an interesting paradox has emerged. In many cases, the industry appears willing to hold AI to standards that it has never consistently applied to itself.

Ironically, the profession's skepticism toward artificial intelligence may produce an unexpected benefit: a higher standard of documentation, transparency and reproducibility across the audit process.

Holding AI to a different standard

When a newly hired associate joins an engagement team, it is understood that they will make mistakes. Despite possessing strong academic credentials, they are often performing procedures for the first time. Instead of expecting perfection, we provide training, supervision, review and coaching. The audit process is designed around the premise that human error is inevitable and quality is achieved through layers of oversight and professional judgment.

The reaction to AI has been fundamentally different. Rather than asking whether controls exist to detect and correct errors, many professionals begin by asking whether the technology can guarantee that errors will never occur. The expectation is often one of near perfection. Any possibility of an incorrect response is viewed not as a manageable risk but as evidence that the technology should not be trusted.

It's an interesting difference because auditing has never been based upon perfection. It has always been based upon obtaining sufficient appropriate evidence and applying a framework of supervision, review and professional judgment to arrive at reasonable conclusions. Human auditors have never been held to a standard of infallibility. Yet that is frequently the standard imposed upon AI.

The discussion becomes even more interesting when viewed through the lens of audit documentation.

Across the industry, there are countless conversations taking place regarding the use of AI for tasks such as document comparisons, sample selection assistance, transaction analysis and workpaper review. In nearly every discussion, the same questions arise: Can the process be reperformed? Can every conclusion be traced back to source data? Can a reviewer determine exactly how a result was generated? Is there a complete audit trail?

These are excellent questions. The irony, however, is they often expose opportunities that extend far beyond artificial intelligence. Consider a sampling decision performed by an experienced auditor. While the final workpaper may document the selected items and the resulting conclusion, the underlying thought process frequently includes years of accumulated experience, professional intuition, risk assessment and judgment. Another experienced auditor may reach the same conclusion, but it is often difficult to fully reconstruct every factor that influenced the original decision.

This reality has traditionally been accepted as part of professional judgment. Yet AI is encouraging us to ask whether more of that process can be documented, structured and reproduced in ways that enhance consistency and transparency.

Transparency is the goal

Audit methodologies were developed to achieve a high level of quality through the effective documentation of conclusions, supporting evidence and significant professional judgments. That framework has served the profession well for decades and remains the foundation of high-quality audits today.

AI presents an opportunity to build upon that strong foundation. As firms explore AI-assisted procedures, they are increasingly asking whether every step can be traced to source data, whether conclusions can be independently reperformed and whether a complete audit trail can be maintained from evidence to outcome. These questions are encouraging the profession to consider how technology might further enhance transparency, consistency and reproducibility.

For example, if an AI-assisted sampling process requires the retention of the source population, the selection criteria, the methodology applied and the resulting output, the engagement team gains an additional layer of visibility into how a conclusion was reached. Similarly, when AI-assisted analysis is supported by a documented record of inputs, assumptions and outputs, reviewers are provided with a clearer understanding of how evidence was evaluated.

Viewed through this lens, the profession's skepticism toward AI may become an impetus for innovation rather than a barrier to adoption. The questions being raised are not challenges to the quality of current audit methodologies. Instead, they reflect the profession's longstanding commitment to continuous improvement and its willingness to embrace new tools that can further strengthen audit quality.

Professional judgment will remain at the center of auditing. No technology can replace an auditor's ability to understand risk, challenge assumptions or evaluate the sufficiency of evidence. What AI can do is help to better capture, organize and communicate that judgment. In many respects, AI is acting as a catalyst, pushing the profession to think more deliberately about how conclusions are formed, documented and supported.

The opportunity beneath the trust debate

The debate surrounding AI is often framed as a question of trust: Can we trust the technology? Can we rely on its conclusions? Can we defend its output?

Those questions are important, but they may not be the most important outcome of the discussion.

Challenging AI to explain itself, document its reasoning and create a clear connection between evidence and conclusions forces the industry to think more deeply about how audit work is performed. 

For generations, the profession has advanced by adopting new tools that increased both efficiency and quality. Electronic workpapers, data analytics and continuous auditing each expanded what auditors could accomplish. AI may represent the next step in that progression.

What makes this moment unique is that AI is doing more than helping perform audit procedures faster. It is encouraging the industry to examine how professional judgment is applied, how conclusions are supported and how audit evidence can be connected more clearly to the decisions being made.

This is why the AI double standard may ultimately be a positive development. The skepticism is healthy, and the tough questions are necessary. The demand for transparency, reproducibility and accountability is exactly what the profession should expect from any new tool.

Yet the benefits of those questions may extend far beyond AI itself.

Our effort to determine how AI can strengthen auditing may lead to new ways to strengthen ourselves. The industry may develop methodologies that are easier to review, conclusions that are easier to defend, and audit evidence that is more accessible and more meaningful. We may find that the same standards we sought to impose on technology become a catalyst for elevating the quality of the audit process as a whole.

If this happens, the lasting impact of AI will not be that auditors worked faster. It will be that auditors became better.