Report details how states can implement reliable, effective generative AI
sankai/Getty Images
For generative AI initiatives to be successful, a new report suggests a slow and steady approach to the technology.
A new report highlights how generative AI could impact state operations and offers insights to effectively implement the burgeoning technology.
The explosive debut of generative AI following the 2022 launch of the popular ChatGPT chatbot catapulted the technology into the public sector’s eyes, and state officials have since explored its potential to streamline workflows and boost productivity.
Over the next few years, generative AI is expected to take over routine government functions — such as program eligibility determination or policy assessments — and improve employee and customer experience, according to a report released this week from the National Association of State Chief Information Officers.
CIOs also anticipate that restrictions against the use of generative AI tools will decline as it becomes more useful for enhancing government processes and functions. Additionally, CIOs reported that states will invest more in data projects to improve the quality and reliability of generative AI initiatives.
But the NASCIO report underscored the importance of an incremental, regulated approach to generative AI before states rush into using it.
The report suggested that states should focus on data quality and accessibility to ensure that generative AI systems are effective and accurate. CIOs, for instance, could collaborate with chief data officers to develop data governance and retention policies to make sure data is clean and reliable for use.
Indeed, state tech leaders underscored that having quality data available for generative AI is a major challenge as governments embark on innovative initiatives using the technology, Route Fifty previously reported.
Another way CIOs can prepare for generative AI implementation is to partner with other tech leaders, like chief information security officers, and academic institutions to develop responsible AI practices and policies that help inform experimentation and early use cases of generative AI.
“Start with small, low-risk pilot projects to identify potential use cases and build on lessons learned,” the report stated. “This type of phased approach provides valuable insights that can inform broader AI strategies in the future.”
Transcribing meetings, managing and generating documents and generating software code, for instance, were the top processes that state CIOs reported using generative AI for currently, the report found. But NASCIO highlighted that states are exploring ways that the technology could be leveraged for other business areas, like education and human services.
Utah, for instance, is exploring how generative AI could assist staff at the state’s tax commission call center. According to the report, authorities plan to use information from tax laws, tax websites and training manuals to develop a large language model that will provide tax agents with “more contextually rich, fact-based information” to support their work, said Utah’s Chief Information Officer Alan Fuller.
The NASCIO report also recommended that states be transparent about the limitations of their generative AI projects, cautioning against getting caught up in the hype of the technology. Asking vendors to disclose how AI is used in their services and products is another way states can manage the rollout of generative AI, according to the report.
“Over the next months and years, we will continue to see creative use cases as generative AI improves, and states become more comfortable with the technology’s ability to provide accurate and trustworthy outcomes,” the report stated. “One thing is for sure, [generative AI] is here to stay.”
NEXT STORY: Charm City bets on blockchain to reduce vacant properties