Leaders in the tech industry reveal how to make A.I. projects a success.

Companies are eager to start artificial intelligence projects, but they must ensure that these cutting-edge initiatives help them increase sales or minimize costs.
That was one of the takeaways from a Fortune Brainstorm A.I. online event earlier this week on using data to better organizational operations. Experts stressed the need of company data scientists or managers educating CEOs about their AI projects and outlining how those efforts might help the company financially.
As Fortune has reported, technologists and corporate finance teams are frequently at odds because the tech side finds it difficult to communicate the advantages of artificial intelligence to accounting teams. Machine learning, for example, may be a useful tool for boosting a company’s IT credentials while also enhancing the speed of a corporate app. However, it might be difficult to link such benefits to a specific financial outcome, such as increased sales or lower operating costs.
Joe Depa, global managing director and data-led transformation lead at Accenture, noted that early engagement with the C-suite, such as chief financial officers, is critical to a successful A.I. project. Executives will be able to keep the project on track in this manner.
“You got to make sure that there are use cases that are driving revenue or cost savings along the way so they continue to have the C-level sponsorship that you need to make these programs successful,” Depa said.
Wells Fargo’s head of consumer data and interaction platforms, Sandra Nudelman, highlighted that her organization is looking to employ artificial intelligence in specific ways to cut costs, such as eliminating credit card theft. However, spending enough time cleaning and prepping the relevant data at the start of an A.I. project is a huge roadblock for businesses. She claims that doing so creates a “foundational layer” that allows organizations to establish new A.I. initiatives more quickly and for less money.
“If you already have your data set up relatively well, it’s relatively easy to pilot and not a whole lot of money out of pocket to do it.”
Bonnie Titone, the chief information officer of Duke Energy, said that her company has a set method for evaluating A.I. projects so they don’t waste money. It allows for eight-week experiments to determine whether an imitative has potential, and if not, the company shelves the project and goes on to the next one.
“I think it’s a safe-to-try model that also helps with innovation so that you’re not squashing ideas, because you don’t have a full business case and you don’t want to go ask for millions of dollars in funding,” Titone said.
Even if an effort only slightly improves operations, a successful A.I. project can result in significant financial gains for a company. Susan Doniz, the Boeing Company’s chief information officer, emphasized that even a small increase in the accuracy of a machine-learning model might have a significant cost impact. Because 25% of an airline’s operating expenditures are fuel-related, a 2% improvement in fuel use accomplished by a machine-learning model corresponds to a “huge cost savings to the bottom line.”
“And also it helps with [our] sustainability agenda as well,” Doniz added. “More fuel optimization also means less fuel emissions as well.”

Based on Jonathan Vanian’s from Fortune