AI is a credibility problem for capital markets

PUBLIC discussion about artificial intelligence (AI) and accountancy has largely revolved around one question: will accountants be displaced?

That is the wrong question.

A more revealing lesson came in 2024, when India’s two main stock exchanges warned investors about deepfake videos impersonating their chief executives and offering fake stock advice.

The episode showed how easily AI-generated narratives can borrow the appearance of authority and enter the market’s information stream. The central issue was not automation, but credibility.

That is why AI is not mainly a labour problem for accountancy. It is a credibility problem for capital markets. When analysis, commentary and even apparent executive communication become cheap to fabricate, markets need firmer anchors inside firms: disciplined reporting, disclosures that can be challenged and accountable ownership of judgment.

In an AI-rich economy, accountancy matters not just as a reporting function but also as part of the infrastructure that makes economic information reliable enough for others to act on. Seen this way, the future of the profession looks different.

For years, accountancy has been seen, sometimes unfairly, as a discipline centred on compliance and historical record-keeping. But its deeper economic role is to establish what is real, what can be verified, which assumptions sit behind reported numbers, and who is accountable when judgment enters the process.

That function becomes more valuable, not less, when AI is widely used. If machines can generate reports, reconcile accounts, draft analyses and flag anomalies at scale, the central question shifts from “can this be produced?” to “can this be trusted?”

That is a much harder question, because credibility in AI-assisted work does not mean proving every model output “true”. In many cases, that is impossible.

It means something more practical: the provenance of the data can be traced; the purpose and limits of the model are understood; material assumptions are identified; human challenge is built into the process; overrides are logged; and someone accountable can explain why the final number or narrative deserves reliance despite uncertainty.

Take three areas where firms are likely to lean more heavily on AI: management commentary, provisioning and disclosure.

An AI-generated earnings narrative may be polished yet subtly overstate what the numbers support. A provisioning model may look more precise while resting on unstable assumptions or poorly governed data. A draft disclosure may sound complete while blurring where human judgment ended and model suggestion began.

In each case, productivity rises. But unless governance keeps pace, so does fragility.

Singapore is not starting from zero. The Monetary Authority of Singapore has moved from broad principles to more testable AI governance through Principles to promote Fairness, Ethics, Accountability and Transparency (Feat) and Veritas.

AI Verify similarly emphasises technical testing and process checks. The Institute of Singapore Chartered Accountants (ISCA) has issued a Responsible Artificial Intelligence Framework in Accountancy.

The strategic opportunity now is to turn these frameworks into an investor-visible market norm: AI use that is reviewable by boards, explainable to auditors, and reflected in disclosures that investors can actually interpret.

Three things follow from this.

First, boards should treat AI as part of the firm’s control architecture, not merely as a productivity tool.

If a bank uses AI to support expected-credit-loss estimates, or a listed company uses it to shape management commentary, the audit committee should not ask only how much time was saved.

It should ask what data were used, which assumptions materially changed, what the model is known to do badly, how outputs were challenged, where overrides were logged, and who ultimately owns the judgment.

Governance has to move upstream, closer to how numbers and narratives are produced.

Second, regulators and assurance providers need to be clearer about boundaries. AI should be integrated into internal controls over financial reporting where it affects numbers or disclosures. But accountancy cannot make forecasts or generative narratives inherently true.

What it can do is strengthen provenance, challengeability, disclosure discipline and accountable ownership. That is the realistic scope of assurance: not certifying machine intelligence, but giving markets confidence that important outputs were produced through reviewable processes with defensible controls.

Third, education and professional development should not collapse AI preparation into software training. Tool familiarity matters, but it is perishable. More durable is the ability to interrogate evidence, understand causality, identify control weaknesses, communicate uncertainty and exercise judgment where responsibility cannot be outsourced to a machine.

The useful accountant of the future will not just know how to use AI. The more valuable one will know when not to trust it, how to test it, and how to explain that decision to others.

This matters, especially for Singapore. A financial centre competes not only on speed and efficiency, but on confidence. If AI lowers the cost of producing financial claims, then trusted governance, credible disclosure and robust assurance become more important advantages.

Trust, in other words, is not a soft virtue. It is economic infrastructure.

So, the future of accountancy should not be framed too narrowly as a struggle for professional survival. The bigger risk is not simply that jobs change. It is that markets become saturated with low-cost intelligence but short on accountability.

The opportunity, conversely, is that accountancy becomes more central, not less: not as a back-office reporting function, but as part of the architecture that makes AI-compatible markets governable.

The real question, then, is not whether accountants will survive AI, but whether a financial centre such as Singapore can preserve market confidence when intelligence becomes abundant but accountability does not automatically keep pace.

It cannot do so without the discipline that accountancy brings.

The writer is the Lee Kong Chian professor of accounting and dean of the school of accountancy at Singapore Management University