Just half of state CIOs say employees use generative AI in daily work, NASCIO report says

Just over half of chief information officers surveyed by the National Association of State CIOs said employees in their organizations use generative artificial tools in their daily work, according to a new report published Tuesday.

The report shows that only 53% of CIO respondents said their employees use AI to perform work tasks — a lower-than-expected number, and one that NASCIO said “doesn’t seem plausible” with AI tools becoming more accessible.

The findings published in the report — which were taken from the generative AI section of NASCIO’s 2024 State CIO Survey and additional interviews with 11 state CIOs — build a more in-depth picture of how state organizations are using AI in daily operations.

NASCIO attributed a large part of the discrepancy to the fact that there are states where generative AI websites or tools are prohibited, though some CIOs said they know employees are using them on personal devices or using other workarounds.

“Our agency blocked access to most free GenAI tools, but we know that employees are still accessing sites and using workarounds. Others are certainly using GenAI that is built into already purchased products,” one CIO said in the report.

The survey also found that 29% of state CIOs said their employees don’t use generative AI at all, 6% don’t know if their employees were using the tech, and another 12% selected “other,” which included the response that the tech being used in limited ways.

For those states where employees are permitted to use generative AI, free online tools such as ChatGPT were the most common, with 65% of CIOs in those states saying they were in use. Another 59% said their organizations were using commercial off-the-shelf AI tools, and 35% of CIOs in those states reported that they are using custom tools developed by a third party or vendor.

As for how the tools are being used, state CIOs said generative AI’s top use cases in government included virtual meeting assistants and transcription; cybersecurity operations; document generation and management; and software code generation. For those who are in the pilot stage with the technology, document generation and management was also top use case, followed by data and predictive analytics, virtual meeting assistants and transcription, and software code generation.

Only 16% of states said they used generative AI tools developed in-house, and 8% of respondents said their state’s executive branch did not use generative AI tools at all.

The NASCIO report also asked CIOs what regulatory actions for generative AI have been implemented in their states, such as the creation of an AI advisory committee and task force, with 78% indicating their state has. Another 72% said their state has implemented enterprise policies and procedures on development or use, and 67% said their state has developed flexible guardrails regulating the responsible use, security and ethics of AI.

Most states also completed an AI inventory, or at least documented use cases in agencies and state applications. However, less than half implemented transparency and accountability rules, codified procurement terms and contract provisions for AI tools, or formalized the structure of data governance for AI.

As for the challenges that state CIOs are facing when expanding the use of generative AI, several CIOs noted issues with reliability, trust, high costs and a need for better governance and training. Other challenges referenced include cultural aspects, like managing the fear of change and encouraging the exploration of the tools, the report said.

“Ultimately, it’s change management. There’s a fear of change and there’s a weakness to identify uses. People say: ‘Oh, I could do that with AI?! I didn’t know AI could help with this.’ It’s getting the word out,” Massachusetts CIO Jason Snyder said in the report.