Only 13 percent of corporate CIOs said they achieved major results with AI, a survey found.

A recent survey by IT market research firm Gartner showed AI use cases have increased rapidly over the past 2 years, but the odds of success remain about 1 in 5.

Gartner shared the findings after surveying 2,002 CIOs and technology executives from May to July this year.

Alicia Mullery (알리시아 멀러리), a vice president analyst at Gartner, pointed to the issue at the Gartner IT Symposium/Expo in Orlando, Florida. "Quantity is not value," she said. She said organisations without fundamentals such as defining the value of a use case in advance and setting criteria for when to stop are paying a price with AI.

Gartner classified CIOs who deliver business results and whose CEOs highly recognise AI value as a "leading group". Organisations that set criteria for not moving AI pilot projects into real operations were 2.3 times more likely to be in the leading group. Those with criteria for shutting down low-performing AI initiatives already in operation were 2.6 times more likely. The share of CIOs with those criteria was 31 percent and 17 percent, respectively.

Cost pressures are also rising. Some 58 percent of CIOs said pressure has grown to cut costs using AI. Even so, 51 percent expected AI to increase total cost of ownership for technology. Mullery said costs for existing technologies are generally predictable, but AI costs are determined probabilistically when it is actually used. She said CIOs need a new approach to funding and management, and to cost accountability.

Workforce redesign was also cited as a task. Organisations that made plans to restructure roles so that people and AI agents work together were 2.4 times more likely to be in the leading group. But only 4 percent of CIOs had such plans.

Gartner said AI is lowering technology development costs while demand for technology work is instead increasing. Some 40 percent of respondents said they plan to increase IT headcount next year. Many CIOs said they would add staff in cyber security and data and analytics. Mullery said role boundaries are blurring, with designers taking on testing and testers joining design. She said the number of roles organisations need will fall, and what each person does will broaden.