ORLANDO, Fla. — All enterprise tech executives are facing the same, expensive balancing act — weighing the risk of AI against its business value, analysts said during the opening keynote of the Gartner IT Symposium/Xpo Monday.
The scales are currently tipped, Distinguished Analysts Kristin Moyer and Daryl Plummer told attendees, as 86% of CIOs see risk growing faster than value.
The stakes, and the price tag, have never been higher. Tech leaders are spending 18 times more on AI than they did two years ago, according to Gartner research. They’re also fending off security concerns as AI’s autonomy grows.
“When it comes to AI, 1,000 questions come to mind. Everything from, ‘will AI help us live to 150 years?’ to ‘will it kill us all in the next 10?’” Moyer said. “It’s like you’re holding two competing ideas in your head at the same time.”
While risk is unavoidable in IT transformations, CIOs need to align risk with the value the technology is creating in order to get the most from their investments. They should also establish shared frameworks for decision-making within their organizations, the analysts said.
Question AI use cases, spending
ROI is lagging, as one in five AI projects are currently succeeding, the same rate as last year, according to Gartner. Nearly 60% of executives say they get positive ROI from an AI project, but the ones that succeed are driving the bulk of AI value at organizations, according to Gartner research.
AI use cases have increased 400% in the last year, but value, not volume, is what CIOs should be aiming for, Plummer said. CIOs should frequently audit their use-cases and be more critical of how and when the technology is being deployed in their organizations.
AI vendors, eager to meet the demand for more services, are set to spend nearly $7 trillion on AI infrastructure by 2030. They push products as a boon for productivity and efficiency, but just 16% of enterprises report productivity gains, Plummer said.
“Vendors want you to use AI agents everywhere, and they want you to build momentum for their products, or to just buy more tokens,” Plummer said. “Instead of monetizing your value or getting your returns, vendors want you to grow theirs.”
Many organizations are using — and paying for — AI tools when simple searches or queries are adequate.
About four in 10 workers said they’ve received AI-generated slop, or low-quality output in the last month, and each occurrence took nearly two hours to clean. For an average Fortune 500 company, that time costs about $52 million a year, Moyer said.
Instead of adding new use cases, Plummer told CIOs to scale the successful ones they already have.
“Stop careless consumption and start outcome-based development,” he said. “You must reduce waste before cutting investment and invest in where outcomes matter most.”
Governance and control
In tandem with auditing AI use-cases, CIOs and the C-suite need to gain control and governance over the technology they’re deploying. According to a recent survey, only 21% of CIOs have governance in place for agentic AI, but 71% plan to deploy it this year.
Shadow AI and dark AI — any AI use that causes harm, whether intended or unintended — persists across enterprises. Risk has only increased as technology becomes more autonomous, and many CIOs don’t know the full extent of the agent use or access in their organizations.
Many industry leaders are now calling for global alignment on guardrails. It’s unlikely, Moyer said, but enterprises can take a similar approach with unified governance across the organization.
In the past, most technology had a visible trail of accountability, but with AI, most do not see harm that was done until after it happens. CIOs should work to create explicit processes and security reviews, as well as strengthen controls surrounding AI use, Plummer said.
There’s no consensus around a model for clear ownership of AI, but CIOs report feeling personal responsibility to manage the technology’s outcomes. Governance can and should sit between the CIO and CFO with regular consultations from CISOs, legal teams and others, Plummer said.
CIOs can begin this process with a few things, including investing in an AI gateway that routes requests to the best model or platform for the test. They should also establish who owns the platforms, tools and models in their organization.
Lastly, CIOs should revisit the stability of their security processes “forever,” Moyer said.
“Without a deep review and audit of your security policies, you’re just asking for trouble,” she added.
