Noel Corriveau, legal counsel for INQ Law of Canada and former senior adviser for AI policy and implementation to the Treasury Board of Canada Secretariat, addressed technology officials from the California Department of Technology (CDT), the Department of Forestry and Fire Protection and other state entities Monday at the third statewide AI Community of Practice discussion. At the event, convened virtually and led by CDT, Corriveau recommended that those present be very mindful of how they operate when moving forward with AI. Among the takeaways:
- Focus on “not the immediate,” Corriveau said, meaning “not to look at the immediate problem but really keep an eye on what’s coming up.” It’s important to look at decision systems to ensure they can capture nuances — but governments must be wary of becoming too focused on the here and now, particularly given the fast pace of AI itself.
“It’s a very hard state to regulate and it’s changing very quickly, and you don’t want to be in a position where you’re just stuck and you’re in the process of trying to amend something” that may already be changing, he said. - Consider how often you refresh your policies around AI. Major technology documents and policy pieces at the state level often don’t refresh yearly. For example, “Vision 2023,” California’s statewide technology strategic plan released in January, followed “Vision 2020,” the strategic plan released in November 2017. Three-year cycles are traditional, Corriveau said — but for fast-moving AI, he suggested streamlining the refresh cycle and potentially refreshing more often, pointing out that in some cases, quarterly refreshes have been weighed.
- Data quality is key to AI, and so is testing. Corriveau reminded those assembled online to be mindful of the required quality of data sets that they’ll need when deploying certain AI systems. Consider, he said, “how can you create a more standardized and consistent way of testing and reporting on your data sets?”
- The Community of Practice won’t reconvene until June, and in the interim, Pam Haase, CDT’s chief of statewide technology policy, directed attendees to a Community Portal. Registration is required to comment or to post content, but content is viewable to all visitors, including posts on AI, AI governance gap analysis, machine learning and robotic process automation.
- And for government employees only, at 1 p.m. April 15, the Community of Practice will host a discussion between two state technology officials. Ajay Gupta, chief digital transformation officer at the California Department of Motor Vehicles, and Subbarao Mupparaju, chief information officer for the Financial Information System for California, will examine “Lessons Learned from State Government Chatbots.”
