How will AI change audit?
From samples to full populations, from checklists to judgement: what technology changes in audit may be less the speed of the work than its centre of gravity.
For a long time, audit was shaped around one constraint: an auditor could not examine every transaction of a company. Sampling, materiality and a risk-based approach were intelligent ways of managing that constraint. Data analytics and AI are loosening it; the real question is what the audit becomes once the constraint is gone.
From samples to populations
Accounting data is now largely digital and structured. It is technically possible to scan every journal entry of a company for weekend postings, round amounts, unusual account combinations or transactions bunched at period end. I applied this approach in the E-Ledger Analysis Engine. This turns the auditor’s question from “which sample should I pick?” into “which exception should I examine?”
Automating routine work
Work such as agreeing amounts between the statements and the notes, validating totals, reconciliations and document matching can be automated to the extent it can be defined as rules. Automating it does not lower audit quality; on the contrary, it makes systematic the very areas most exposed to human error. TFRS/IFRS Reporting Software is an example for statement–note checks.
What will not change
AI does not deliver the final judgement on whether an accounting estimate is reasonable, whether the substance of a transaction differs from its form, or how credible management’s explanations are. Professional scepticism, understanding context and evaluating evidence will remain the core of the audit.
New tools also bring new risks: explaining why a model flagged what it flagged, the integrity of the data used, and the auditability of the analysis itself.
The auditor’s profile
The most concrete consequence will be in the skills expected of auditors. Accounting and reporting knowledge will remain the foundation; but the value of an auditor who can read data, build simple analyses and understand the limits of the tools will rise. Technology is less likely to replace the auditor than to widen the gap between a good auditor and an average one.
I discuss the in-house counterpart of audit — financial control — on the audit and financial control page.