CFOs Add AI Fact-Checking to Duty List
A new Datarails report suggests many finance leaders are spending substantial time checking AI outputs as pressure to expand adoption grows.
Sign up for smart news, insights, and analysis on the biggest financial stories of the day.
Welcome to the fact-checking and editing world, CFOs.
A September report from FP&A technology provider Datarails, based on a survey conducted by Global Surveyz Research, found that 96% of the 270 CFOs and other finance leaders spend at least 10% of their workday verifying or correcting finance AI outputs.
For roughly a third of CFOs, that work can consume up to half of the day. Sounds fun, right?
Conducted in July, the report covered finance leaders at companies with at least 1,000 employees and $100 million in annual revenue. The findings also raise questions about how much AI is actually improving productivity.
Their responses suggest that the pressure to move faster with AI comes with new responsibilities and not exactly the payoffs some CFOs are looking for.
Whack-a-Mole
The pressure to implement AI, whether it’s from the board, CEO or an overworked finance team, is on. The survey found 76% of respondents faced high or very high pressure to fully implement AI in finance.
Feeling behind? Don’t worry:
- Only 7% of CFOs said their finance function was fully ready across all workflows from a data and infrastructure standpoint.
- Among finance leaders whose data was spread across disconnected systems, 76% reported spending more than a quarter of their time on AI verification.
- But regardless of the process in which AI is verified, there’s still this finding: 56% said team members had received materially different outputs from the same prompt and data.
“AI was supposed to take the manual work off finance teams. For most of them, it’s added a second job on top: checking the AI,” Jonathan Marciano, vice president of brand and communications at Datarails, told CFO Upside. In finance, he added, “a confident wrong answer is worse than no answer.”
Auditability, unsurprisingly, was the most frequently cited barrier to trusting AI with mission-critical tasks, identified by 75% of respondents. Accuracy and hallucination concerns followed at 71%.
More Work, More Spend
As pressure and workloads grow around AI, so does the desire to increase the amount of capital allocated to these tools. More than half of respondents said they plan to expand AI licenses over the following year, and roughly another third said their organizations “blew through” AI budgets by at least 10% over the previous 12 months.
The appetite for wider access remains substantial, but prepare to keep shelling out cash: The report says finance and accounting software spend in 2025 averaged $361,000.
Bring This to the Board. Here’s a useful question for the next meeting when your board or fellow executives are asking you about progressing automation and AI in your finance function at a pace you’re not so comfortable with: How many employees and how much time will an AI-generated answer require before it becomes usable?