For decades, the junior accountant's first years were built on repetition: vouching transactions, tracing balances, and asking the questions that a more experienced colleague would later refine. That apprenticeship model is now under pressure. A study released by Chartered Accountants Worldwide and Chartered Accountants Ireland, reported by Business Eye, suggests that AI has moved from experiment to habit, and a parallel KPMG pilot hints at what that habit could mean for audit quality and for the people who learn the trade by doing it.
The Numbers Behind the Headline
The headline finding is striking: roughly eight in ten chartered accountants now use AI weekly. The study also found that 85% of accountants feel AI is enhancing their day-to-day workflow. Usage and sentiment are related but not identical. The first measures behaviour, the second measures perception, and both matter for firm leaders deciding where to invest.
| Indicator | What It Shows | Implication for Practice |
|---|---|---|
| Weekly AI use (approx. 8 in 10) | AI is embedded in routine professional work | Usage policies and data controls are now operational necessities, not future planning |
| 85% report workflow enhancement | Practitioners see tangible productivity benefits | Adoption is driven by demonstrated value, which raises the stakes for quality assurance |
| KPMG pilot: 100+ AI agents run alongside human auditors | AI tested on a listed firm's financial statements in parallel with the audit team | Audit methodology and evidence standards will need to address AI-generated findings |
| Senior leaders: human intervention still required | Regulatory adherence cannot be fully delegated | Professional judgement remains the accountable element |
"85% of accountants feel AI is enhancing their day-to-day workflow." The finding frames AI as a productivity tool that practitioners already trust, which makes governance and training the next frontier.
The Questioning Edge: What the KPMG Pilot Shows
The more revealing evidence comes from a pilot audit reported by the Irish Times. KPMG ran more than 100 AI agents in parallel with human auditors on a listed company's financial statements. The headline result was that the technology asked better questions, frequently probing issues more deeply than junior staff did.
Where AI Added Value
The finding is not that AI replaces judgement. It is that AI can be relentless in interrogating the same data from multiple angles, without fatigue or the social pressure that can cause a junior team member to accept an explanation too quickly. In an audit, that kind of persistence is valuable, particularly in areas such as unusual journal entries, reconciliations and estimate-heavy balances.
AI "asked 'better questions'" than KPMG accountants in a pilot audit of effectiveness, according to the Irish Times coverage of the KPMG test.
Where Humans Remained Essential
Senior leaders in the pilot were clear that human intervention remained necessary to ensure regulatory adherence. Questions are not the same as conclusions. An AI agent can identify an anomaly; it cannot take professional responsibility for whether the financial statements give a true and fair view, or whether the evidence meets the standards that regulators and the courts will later examine.
The Training Dilemma for Irish Practices
If AI handles much of the first-pass questioning, the traditional route into the profession is at risk of hollowing out. Junior staff have historically learned to ask good questions by working through large volumes of routine evidence. Firms that automate that layer without redesigning development pathways may find that their future seniors lack the depth they need.
Several implications follow for Irish practices:
- Redesign the apprenticeship: Juniors should review AI-generated findings and explain why they agree or disagree, turning automated output into a learning exercise.
- Teach scepticism explicitly: Training should cover how to challenge a model's reasoning, including where its training data or inputs may be incomplete.
- Protect exposure to judgement: Rotations should ensure junior staff still encounter complex estimates, client conversations and regulatory queries, not only AI-supported checklists.
- Measure competence, not only output: Performance reviews should assess the quality of questions asked, not just the volume of work completed.
Certification and Ethical Deployment
The Chartered Accountants Worldwide and Chartered Accountants Ireland study also noted that the professional body is introducing new AI certification programmes. These aim to support members and Irish businesses in deploying the technology ethically. For practitioners, the value of such programmes will depend on how they are structured. Useful certification will likely address:
- Data protection and confidentiality when client information is processed by external AI tools
- Documentation standards for AI-assisted working papers and audit evidence
- Model limitations, bias and the risk of plausible but incorrect outputs
- Accountability frameworks showing where the professional signs off and where the tool is advisory only
Certification alone will not close the gap. It needs to be embedded in firm procedures and supported by partners who model the behaviour they expect.
Practical Steps for the Next Twelve Months
For practice leaders in Ireland, the sequence of actions matters. A workable approach might look like this:
- Audit your current AI use: Map which tools staff use weekly, which client data they touch, and where outputs enter working papers.
- Set a written review standard: Require documented human review for any AI-generated finding that influences an opinion, a conclusion or a client recommendation.
- Pilot parallel testing: Following the KPMG approach on a smaller scale, run AI alongside experienced staff on a selected engagement and compare the questions raised.
- Rebuild training curricula: Update graduate and trainee programmes so that AI supports learning rather than replacing the struggle that builds judgement.
- Engage with certification early: Ensure partners and managers complete relevant AI programmes before juniors are expected to use tools unsupervised.
Looking Ahead
The weekly use figures and the KPMG pilot point in the same direction. AI is becoming a standard part of accounting work, and its strongest contribution so far is in asking more persistent and more probing questions of the data. That is a genuine benefit, but it shifts the professional challenge. The scarce skill in the next decade will be judgement: knowing which questions matter, which answers can be relied upon, and who stands behind the final opinion. Irish firms that treat training as the foundation of AI adoption, rather than an afterthought, will be best placed to build that capability before the apprenticeship model is lost.
