Why should corporate finance departments adopt artificial intelligence? Contrary to what many assume, the biggest gains are not in efficiency, as represented by faster closings, reduced errors and cost savings.
Such traditional measurements of ROI don’t matter nearly as much as AI’s capacity to improve companies’ judgment, according to KPMG, which recently surveyed 1,013 senior finance leaders across 20 countries who were working at organizations with at least $250 million in annual revenues.
In the aggregate, 70% of survey respondents reported that over the previous 12 months AI had moderately or significantly improved decision-making quality, while 71% said the same for decision speed. And nearly as many (64%) said AI improved forecasting accuracy.
“Judgment-heavy work carries the most accumulated weakness in the finance function,” KPMG wrote in its research report. “It runs on inconsistent data, under-invested tooling, and manual judgment built into the numbers. AI has more to gain here.”
The report added that traditional processes are improving too, but at smaller margins.
Overall, 75% of the respondents reported active AI usage across finance, compared with 30% in 2024. Meaningfully, 71% said such usage has met or exceeded expectations. But only 23% of the finance leaders were in the “exceeded” bucket, which KPMG said suggests AI adoption is moving faster than companies’ ability to translate it into enterprise-wide performance gains at scale.
Companies that formally track AI-related KPIs are more successful at reaping such broad and deep gains, the research showed. On average, 68% of them reported moderately or significantly improved ROI, compared with 58% of non-trackers. “The pace of change makes confidence in AI-generated conclusions essential,” said Steve Chase, global head of AI and digital innovation for KPMG International, in the report.
However, only 29% of the respondents said their companies formally track where AI adoption fails. They can see what AI is delivering, but not “why it breaks or where it is exposed,” KPMG wrote.
Indeed, assurance readiness — the extent to which a company ensures that its data, processes, internal controls and systems are robust enough to withstand a formal third-party audit — is on its way to becoming “table stakes” with respect to AI outcomes.
That’s particularly true at the top end of the performance curve. For example, 60% of companies KPMG identified as agentic AI leaders are strongly assurance-ready for AI-enabled finance processes, compared with 42% of all companies.
But KPMG pointed out that the 60% figure among leaders shows that even the most advanced organizations are still building this infrastructure.
