CFOs are Overestimating AI Usage

CFOs are overestimating AI usage in many organizations because executive perceptions often outpace how extensively employees actually use AI in their daily work. While finance leaders increasingly view their companies as AI-enabled, many teams are still in the early stages of adoption, with limited integration into core FP&A, forecasting, reporting, and decision-making processes. Closing this gap requires more than investing in AI technology. It demands accurate measurement of AI adoption, employee training, strong data governance, and workflows that embed AI into everyday finance operations rather than treating it as an occasional productivity tool.

The AI Confidence Gap

More than a third of CFOs and senior finance leaders consider their organization’s AI capabilities “leading.” Ask the managers running day-to-day operations, and a much smaller share agrees. This disconnect sits at the center of a growing problem: CFOs are overestimating AI usage, and the gap between perception and reality is starting to show up in missed growth targets, misallocated budgets, and strategic decisions built on capabilities that don’t yet exist on the ground.

Recent research into AI maturity in finance suggests this isn’t a minor miscalibration. It’s a consistent pattern, and it’s an expensive one. Understanding why senior leaders consistently overstate their AI maturity, and what to do about it, starts with looking at where executive perception and operational reality part ways.

Across multiple studies of finance organizations, seniority correlates directly with a rosier view of AI progress. Senior leaders are far more likely than their teams to rate the company’s AI maturity as advanced, report fewer AI adoption barriers, and plan larger AI budgets. Wharton School researchers found a similar split when surveying leaders at companies earning more than $50 million in annual revenue:

• 45% of executives reported strong ROI from AI investments;

• Compared to just 27% of middle managers.

• Asked whether their company was adopting AI faster than competitors, 56% of executives said yes compared to only 28% of managers.

This is the AI confidence gap in action: two groups inside the same organization, working from two different pictures of what’s actually working. 

AI Maturity Versus AI Adoption

Part of the confusion comes from conflating AI adoption with AI maturity. Adoption measures whether a tool is being used. Maturity measures whether that tool has changed how work actually gets done. A company can show widespread AI adoption, frequent logins, dozens of licenses, AI features switched on across departments, while remaining low in maturity because none of that usage has improved decision speed, forecast accuracy, or planning confidence. The distance between perceived AI maturity and actual AI adoption is exactly where most of this confusion lives. CFOs often point to adoption metrics as proof of maturity, but the two move at very different speeds, and treating AI maturity versus AI adoption as the same measurement is where the overestimation tends to begin.

Why CFOs are Overestimating AI Usage

Several forces push senior leaders toward an inflated view of their own AI maturity, and most of them are structural rather than a matter of ego.

The Urgency-Confidence Paradox

Senior finance leaders report higher levels of uncertainty about the external business environment than the managers below them, yet they’re also the most confident about their organization’s internal AI readiness. That pairing makes sense once you consider the incentive: the more uncertain the outlook, the stronger the pull to point to AI investment as evidence of preparedness. But urgency isn’t the same as readiness, and conflating the two is one of the clearest reasons why CFOs overestimate AI maturity rather than measure it.

Different Tasks, Different Realities

Executives and managers also tend to use AI differently. Senior leaders typically apply it to high-level synthesis, strategic drafting, and decision support, tasks where today’s tools perform reliably well.

Managers deploy AI inside messier, higher-stakes workflows:

• Legacy processes

• Teams with uneven technical comfort

• Outputs that have to be consistently accurate

When a tool underperforms, it’s usually the manager’s team that absorbs the fallout, not the executive who approved the investment. That structural gap in exposure is a major reason why finance leaders misjudge AI progress so consistently, because they’re seeing a narrower, easier slice of the workflow than the people actually running it.

How AI Maturity Affects Business Performance

The gap matters because AI maturity is closely tied to financial outcomes. Companies at the most mature stage of AI adoption have grown revenue at roughly three times the rate of early-stage organizations, and they’re the only group whose actual growth has outpaced their own projections. Every other maturity tier has fallen short of what it forecasted. The performance bump shows up disproportionately late in the maturity curve. The quality of decision-making climbs sharply between advanced and leading organizations, and confidence in planning outcomes rises even more steeply between developing and leading stages. This is the AI planning advantage that comes with genuine maturity: not a smooth, linear payoff, but a threshold effect that rewards companies that can accurately identify which side of that threshold they’re actually on.

Signs Your Company Is Not AI-Mature

A few warning signs tend to show up consistently among finance teams that are earlier on the maturity curve than their leadership believes:

• AI tools are used inconsistently across teams, with usage driven by individual habit rather than an embedded process.

• Forecast and planning outcomes haven’t measurably improved despite a tool rollout.

• Data quality issues, fragmented systems, or manual reconciliation still bottleneck AI-driven workflows.

• Frontline staff describe AI as “helpful sometimes” rather than central to how they work.

• Leadership cites adoption numbers — logins, licenses, usage rates — as evidence of progress, with no metric tied to decision quality or speed.

These signs point to a deeper issue: weak data readiness for AI. Without clean, connected, and well-governed data, even widespread tool usage stalls before it produces the planning or forecasting gains that define true maturity.

How to Measure AI Maturity

Closing the confidence gap starts with replacing assumptions with structured measurement.

Build Upward Feedback Loops

Wharton researchers describe “upward feedback loops” as a precondition for meaningful AI transformation: structured channels where frontline and managerial staff can report what’s actually working, what’s failing, and where their daily experience diverges from leadership’s expectations. Nearly two-thirds of executives in their study said they had grown “much more positive” about generative AI over the past year, compared to only 39% of middle managers. Without a formal mechanism to surface that gap, it stays invisible until it shows up in a missed target.

Use a Structured AI Maturity Assessment

Self-reported confidence is an unreliable gauge of AI adoption maturity. A more reliable approach weighs leadership’s perception against ground-level data: how forecasts perform against actuals, how often manual workarounds are still required, and how consistently tools are used across teams rather than by a handful of power users. Running this kind of structured AI maturity assessment, even informally through a cross-functional audit, gives finance leaders a benchmark grounded in evidence rather than optimism, and makes the AI implementation reality visible before it’s reflected in a missed forecast.

How to Improve AI Maturity

Improving AI maturity isn’t primarily about adding more tools; it’s about closing the distance between what leadership believes is happening and what’s actually happening on the ground. That means treating data readiness as a prerequisite for any new AI initiative, building regular channels for managers to report friction points without those reports being read as resistance, and tying AI investment decisions to operational metrics rather than adoption metrics alone. The finance functions seeing the biggest gains from AI aren’t necessarily the ones spending the most. They’re the ones whose leadership has an accurate read on where the organization actually sits on the maturity curve, and who size their next investment to match that reality rather than their hopes for it.