The state’s AI and Analytics Center of Excellence (AI CoE), stood up in June, submitted its initial report on AI to the governor Friday, outlining recommendations on data protection, government workforce training, and formalizing AI governance and a secure AI infrastructure.
“This report outlines a vision and action plan to protect Arkansans and their data while leveraging AI to improve government efficiency, drive economic growth, and prepare our workforce for the jobs of the future,” state Chief Data Officer Robert McGough said in a news release.
Arkansas joins multiple states that are initiating AI oversight with special legislative committees, working groups and centers for excellence. The CoE model has continued to be a trend in state agencies to enable AI while ensuring its ethical use. South Carolina was set to establish its CoE last month, Utah’s example is about six years old, and Texas’ center is entering its fifth year.
Arkansas’s AI CoE report recommends creating a chief AI officer (CAIO) position reporting to McGough that would serve on the state’s Data and Transparency Panel (DTP), which advises on data governance and data sharing. (McGough is chair of both the DTP and the AI CoE.)
Initially, two agency use cases were suggested: unemployment insurance fraud and recidivism reduction. They were identified by available data, funding, state priorities and buy-in, according to a June news release. The new report doesn’t explicitly name what projects are to be piloted, but rather lists potential areas that AI can impact. “Arkansas can drive meaningful improvements,” the report said, by using tools to boost the labor force, reduce recidivism, improve maternal health with data analytics, address food insecurity, attract employers by training AI-ready workers, enhance education and prevent fraud.
All recommendations in the report have safe, responsible AI as their centerpiece. Guardrails include nine principles, which are prioritizing privacy, ensuring transparency, protecting sensitive data, providing explainable decisions, requiring human oversight for critical decisions, assessing accountability prior to deployment, securing infrastructure, ensuring user readiness and preventing misuse or harm to residents.
