If an artificial intelligence tool gives you a recommendation that goes against your own judgment, what do you do? Some finance leaders, apparently, are choosing to follow the recommendation anyway.
In a survey conducted by enterprise planning company Board, nearly half (48%) of the CFOs queried said they’d follow an AI recommendation even if it went against their judgment.
Finance chiefs were more apt to follow such recommendations compared to their fellow C-Suite counterparts, with just 11% of COOs and 33% of CIOs saying they would do the same under the circumstances.
“Those differences have practical implications for cross-functional planning,” Board officials wrote in the report. “A recommendation accepted by finance may receive greater scrutiny from the operations team responsible for putting it into practice.”
Board’s survey, which was conducted in May and June, went to 100 CFOs, 100 CIOs and 100 COOs working at businesses with $100 million or greater in annual revenue. The survey also detected a relatively high level of usage of large language models, tools like ChatGPT and Claude, for decision-making, with 61% of respondents citing such tools “among the external sources that most influence strategic decisions.”
By the numbers
48%
of CFOs follow an AI recommendation even if it “conflicts with their judgment”
33%
of CIOs do the same
11%
of COOs do the same
The findings come as researchers elsewhere continue to probe the longer-term implications of a practice known as “cognitive offloading” brought on by the latest AI tools.
Gordon Pothier, Board’s CFO, said in an interview that his company’s findings show the pressure that C-Suite leaders are under from boards and others to quickly adopt the nascent technology. “There’s an expectation that it’s the smartest person in the room,” he said of artificial intelligence. “It may not be. … It’s smart, but it doesn’t always have the context.”
LLMs were not, of course, the sole influence on executive strategy, per Board’s findings. Forty-two percent of respondents cited industry peers and professional networks among their top three sources influencing strategy, while 30% pointed to “market trends and competitive intelligence.” Just about a quarter (28%) picked external consultants and advisory firms among their top three, while only 5% cited media and thought leadership.
And perhaps unsurprisingly, given how quickly companies are rushing to implement AI tools, governance around such tools remains in flux. Across all respondents, 39% said they have “formal governance and escalation processes for AI-driven decisions.” That number ticked up a bit, to 45%, among the CFO set. But it was just 26% among COOs.
A similar share of operating chiefs (27%) also cited “unclear ownership when an AI-driven decision goes wrong.” The same percentage said that adoption of the tools is “moving faster than teams can handle.”
“What we’re finding is that CFOs are thinking about [governance], but maybe don’t have the right structure in place yet,” Pothier said. He said that many executive teams still want to keep a “human in the loop,” especially for anything involving money or transactions.
As an example, he pointed to some of Board’s own customers in retail and supply chain operations: “They’re moving inventory around. They’re making big decisions based on the information they’re getting from Board. I don’t think you want to do that just through an agent.”
When it comes to technology-influenced decision-making, Board’s report recommended that companies “establish decision rights before AI is used.”
“Executives need to know how an AI recommendation was formed, which assumptions influenced it and where human review is required,” the report stated. “Those expectations should be set before the recommendation reaches a consequential decision.”