The debate over whether artificial intelligence belongs in finance is effectively over. What’s holding adoption back is not ambition, but trust. A new research study from Wakefield Research reveals that while CFOs are eager to deploy AI, they are resisting solutions that force a trade-off between speed and control.

The study surveyed 100 CFOs at mid-market U.S. companies with annual revenues between $50 million and $500 million. Between 60 and 77 per cent of respondents said they plan to adopt AI in finance, depending on the use case. Yet despite this momentum, execution remains constrained by concerns around accuracy, transparency, and auditability.

According to the findings, 96 per cent of CFOs see AI’s biggest benefit as freeing teams to focus on strategic work. However, only 14 per cent fully trust AI to deliver accurate accounting data without oversight, and 97 per cent insist that human supervision remains critical. Rather than a contradiction, the report suggests this reflects a clear vision of how CFOs want AI to operate.

The research highlights dissatisfaction with two prevailing models. AI copilots, often embedded in legacy systems, still require transaction-by-transaction review, delivering only marginal productivity gains. At the other extreme, fully autonomous AI agents promise end-to-end automation but lack verifiable accuracy, audit trails, and contextual understanding—creating unacceptable financial risk.

CFOs instead are calling for what the report terms “intelligent escalation”: AI systems that autonomously process routine transactions but recognize ambiguity and escalate decisions to humans with full contextual insight. One CFO described the ideal solution as “an autopilot—fast, accurate, and with the judgment of our most reliable accountant.”

“The bottleneck isn’t intelligence, it’s judgment,” said Ramnandan Krishnamurthy, co-founder and CEO of Maximor. “As AI capabilities become commoditised, the real differentiator is knowing when to act and when to ask.”

Dominic Rand, CFO of Kiva Brands, echoed the sentiment, noting that many AI tools either introduce opacity or add workload. “What we needed was AI that could handle the routine with precision and escalate when judgment was required,” he said.

The study concludes that finance leaders are no longer asking for smarter AI, but for AI that earns autonomy through transparency, verifiable accuracy, and sound judgment—reshaping the future of AI adoption in finance.