There is a gap at the heart of most companies’ AI transformation stories, and nowhere is it more visible than in the finance function. The tools are available. The investment is flowing. The executive mandate is clear. And yet, in organisation after organisation, AI adoption in finance is producing underwhelming results — not because the technology is inadequate, but because the human and organisational conditions surrounding it are.
That is the central finding of research published in Harvard Business Review by Professor Kristof Stouthuysen and doctoral researcher Angel Oganesian at Vlerick Business School’s Centre for Financial Leadership and Digital Transformation. Drawing on seven years of engagement with CFOs, finance leaders, and business partners across industries — hundreds of executive conversations and research projects — they sought to answer a deceptively simple question: what does it actually take for the finance function to lead, not lag, in the AI era?
Their answer is more nuanced than most AI adoption playbooks acknowledge. AI adoption in finance is indeed widespread. But its impact, the research reveals, is consistently limited by three forces: organisational misalignment, digital overload, and the challenge of integrating new technologies without overwhelming human decision-making.
Adoption Is Easy. Impact Is Hard.
The numbers paint a picture of rapid take-up. AI adoption in finance has surged to 72% of organisations using AI tools in their operations, up from 34% the prior year, according to Protiviti’s Global Finance Trends survey. Gartner estimates that 90% of finance functions will deploy at least one AI-powered solution by the close of 2026. The technology is clearly arriving.
What is not arriving at the same pace is meaningful impact. A Wolters Kluwer survey found that 86% of North American finance teams are still in the early stages of AI adoption — using tools, but not yet scaling results. The L.E.K. Consulting annual CFO survey found that strategic intent “far outpaces operational reality” for most finance organisations. More than 60% of CFOs in a separate Economist Impact study cited upskilling and hiring digitally fluent talent as top challenges, with fragmented systems and limited real-time data access adding further friction.
The Vlerick research gives this gap a name: organisational misalignment. When AI tools are introduced into finance without aligning processes, incentives, and people around them, the tools tend to sit on top of existing workflows rather than transform them. The result is a finance function that is technically more sophisticated but not meaningfully more effective.
The Three Forces Limiting Finance AI
Organisational misalignment starts at the top. When AI adoption is treated as a technology project rather than a strategic transformation, it tends to be led by IT rather than finance leadership. The consequence is tools that solve the problems IT understands — data storage, processing speed, system integration — rather than the problems finance leaders actually need solved: better forecasting, faster close cycles, more useful business partnering. Successful AI adoption in finance requires the CFO to act simultaneously as architect and change agent: designing the technology strategy while also guiding the team through new workflows and building genuine trust in AI-generated outputs.
Digital overload is the second, underappreciated failure mode. There is an assumption that more technology is always better. In practice, finance teams can be buried under dashboards, alerts, reports, and model outputs to the point where signal disappears into noise. What Stouthuysen and Oganesian identify is that integrating AI without managing the cognitive load it creates can actually impair decision-making rather than enhance it. The goal of AI in finance is not to produce more information. It is to produce better decisions. That distinction is consequential.
Overwhelming human judgement is the third limit. AI tools in finance — particularly in forecasting, anomaly detection, and scenario modelling — are only as valuable as the human interpretation layered on top of them. When AI outputs are trusted without scrutiny, errors compound and biases embed. When they are second-guessed constantly, the efficiency gains disappear. The organisations that succeed are those that have invested in what the research calls a “human-in-the-loop” design: machines handle the data-intensive processing while humans bring business context, interpretive judgement, and stakeholder accountability.
What the Leading Finance Functions Are Doing Differently
The research identifies a clear pattern among finance teams that are actually translating AI investment into results.
They start with clear use cases rather than broad mandates. Rather than deploying AI across the finance function simultaneously, leading organisations identify the highest-value, lowest-complexity starting points — typically accounts payable automation, variance analysis, and forecasting — and build confidence before expanding. Ajinomoto Frozen Foods, for instance, reduced budget management time by 88% and achieved same-day P&L reporting after deploying CCH Tagetik’s AI-enabled platform, having first done the foundational work of cleaning data and establishing clear process standards.
They treat data quality as a prerequisite, not an afterthought. Gartner estimates that poor data quality costs companies nearly $12.9 million annually, and by the end of 2026, roughly six in ten AI initiatives will be abandoned because underlying data was not prepared properly. The finance teams succeeding with AI are those that invested in data governance before implementing tools — documenting sources, resolving inconsistencies, and establishing single sources of truth for key financial data elements.
They build AI literacy throughout the team, not just at the top. Workers with meaningful AI skills command a 56% wage premium in the current market. But beyond the financial incentive, finance teams that have invested in broad AI literacy — not converting everyone into data scientists, but helping all staff understand what AI can and cannot do, when to trust outputs, and when to interrogate them — consistently outperform those that treat AI competency as a specialist skill.
They redesign roles rather than simply adding AI to existing ones. The most significant failure mode the Vlerick research identifies is bolting AI tools onto unchanged job designs. Transaction-heavy roles — invoice processing, journal entry posting, reconciliations — are being automated at scale. The finance teams that are capturing the value of that automation are the ones that have deliberately redesigned the roles that remain, shifting focus from data processing to insight generation, from historical reporting to forward-looking business partnering.
Deloitte’s research found that 64% of finance functions plan to infuse more technical skills within their teams through 2025 and 2026. Hewlett Packard Enterprises CFO Marie Myers describes the outcome that model enables: “Democratising AI means making advanced tools accessible and actionable for everyone… These innovations free our teams to focus on higher-value work — partnering with business leaders, shaping strategy, and driving growth.”
The Strategic Imperative
The Vlerick research lands at a moment when the CFO role itself is being redefined. Nearly 90% of CFOs report they are more involved in digital transformation and risk management than three years ago. The finance function is no longer simply the scorekeeper. It is expected to be a proactive partner in strategic decisions, a provider of real-time scenario analysis, and a guardian of organisational resilience in environments of persistent uncertainty.
AI is the enabler of that evolution. But it requires more than tools. It requires organisational alignment, investment in human capability, thoughtful role design, and the wisdom to distinguish between generating more data and making better decisions. As the Vlerick researchers put it, the finance function that wins in the AI era will not be the one that adopted AI first. It will be the one that figured out how to make AI useful — for its people, its processes, and the business it is built to serve.
The technology is ready. The question is whether the organisation around it is.
Written by Shalin Soni, CMA specializing in financial analysis, global markets, and corporate strategy, with hands-on experience in financial planning and analytical decision-making.
ALSO READ
• Warren Buffett’s Most Powerful Lesson: Why the World’s Greatest Investor Says Money Isn’t the Point
• Navy Exchange Competes with Walmart and Amazon to Fund Military Programs
• Tesla Targets Japan’s Top Imported Car Spot with Aggressive EV Expansion
Source: Based on Harvard Business Review and publicly available information.