Most chief audit executives can’t tell you what AI is worth yet
Auditors are using AI in their daily work, and their departments have mostly left them to figure it out alone. 93% of audit leaders and auditors report some level of AI use, while 15% say their department has deployed formal use cases and runs them routinely in audits, according to Gartner.
Chief audit executives (CAE) have to defend that arrangement to stakeholders, and most of them cannot say what it returns. Of the 142 CAEs polled in May, 54% have not started measuring the value of audit’s use of AI. Seven percent tie AI to cost metrics such as reduced external spend or avoided hiring, which is worth remembering the next time someone suggests the tools should pay for themselves in headcount.
Only 38% of the 161 CAEs in that May poll have an AI strategy at any level, and most of those folded it into the overall department strategy instead of writing a separate one for audit. Another 39% are building one now, which suggests the gap is closing.
“AI adoption is widespread across audit teams, but the lack of a formal strategy means most audit functions are not realizing the full potential of these technologies,” said James Bourke, Director Analyst in Gartner’s Risk & Audit Practice.
Where the tools get used
Drafting and planning dominate. Among 743 respondents, 60% use AI for engagement preplanning work such as research and brainstorming risks, and the same share use it to draft audit issues, ratings or reports. Smaller groups apply it to planning deliverables such as risk and control matrices, the grids that map each risk to the control meant to catch it, and to reviewing drafts before they go out.
Audit testing sits at 30%. Testing is the part where an auditor checks whether a control works, and it changes the conclusions in the report, so a hallucination or an omission there costs more than a clumsy paragraph in a draft. Departments using AI for testing need tighter validation of what the model produces, plus documentation that holds up when a regulator or an external auditor asks how the answer was reached.
The barriers are people and process
Audit leaders and auditors named unclear expectations for AI tool use as the most common obstacle, at 48%, followed by variation in how individual auditors use the tools. Technology and tool problems such as model selection and data access ranked below both. Of the people who answered, more pointed at how their colleagues work than at the software.
When everyone uses a tool their own way, the department gets variable output and no basis for comparing one engagement to the next. Structured use cases for high-value workflows are the fix Gartner puts forward, and 12% of respondents are piloting them.
