CIOs are bullish on the possibilities for generative AI: report

CIOs are increasing their overall uptake of generative AI, pushing AI from its current role in isolated pockets of the enterprise into more organization-wide uptake and speeding the adoption of the technology across new industries, a new survey found.

The survey, published today by MIT Technology Review Insights and sponsored by enterprise data management company Databricks, polled 600 “senior data and technology executives.” It predicted that just about every industry will eventually find a use for generative AI in the near future. Retailers could use the technology for scheduling and installation of heavier goods, manufacturers could use it as a virtual “co-pilot” for service and repair technicians, and media outlets could use it to write articles and headlines. (Note: This article was written without AI assistance.)

Furthermore, there’s now more of an expectation that the advent of generative AI will improve existing business AI use cases. Chatbots for customer and employee support, for example, are likely to be improved by wider uptake of generative AI, as well as business transformation efforts around unifying data stores and similar.

CIOs generally think that AI has already proved a wise investment in certain areas of business, most notably the “democratization” of data (referring to internal and external data sharing), at 32% of those polled, and AI-powered business intelligence and analytics, also at 32%. Those CIOs are also bullish on the effect of AI on the workforce, according to the report, viewing the technology as a way to help workers do their jobs, rather than a threat to the existence of those jobs.

AI should be thought of as a co-pilot, rather than a rival, to human workers, according to Matei Zaharia, co-founder and CTO of Databricks, who was quoted in the report.

“I don’t think they’ll be able to do extremely complicated work that requires a lot of planning fully automatically,” he said. “I don’t think there’s anyone whose job is just the super-simple stuff that can be done by a language model.”