California is poised to explore the potential of generative artificial intelligence (GenAI) to address various public service challenges, including traffic congestion and tax guidance, as part of Governor Gavin Newsom's initiative to integrate innovative technologies into government operations.
In a move towards this goal, the state has entered into agreements with five companies to develop GenAI tools utilizing technologies from prominent firms such as OpenAI and Anthropic, backed by industry giants like Microsoft, Google, and Amazon. These tools aim to enhance service delivery to the public across various domains.
Secretary of government operations for California, Amy Tong, expressed optimism about the collaboration, highlighting the companies' commitment to utilizing GenAI for governmental service delivery.
The companies will undergo a six-month internal trial period, during which state workers will test and evaluate the efficacy of the tools. The initial agreement involves a nominal payment of $1 for each proposal. This trial phase allows the state to assess the feasibility of implementing these tools under new contracts, considering the state's budgetary constraints. Importantly, the tools are designed to pose low risk, with no access to confidential or personal data, according to an administration spokesperson.
Governor Newsom has positioned California as a global leader in AI technology, citing the state's prominence with 35 of the top 50 AI companies worldwide. He issued an executive order mandating the exploration of responsible GenAI incorporation last year, with the state initiating the process by inviting technology companies to propose GenAI tools for public services. Last month, California released guidelines on the procurement of such tools by state agencies.
Despite the potential benefits of GenAI in enhancing government efficiency, concerns regarding oversight and risk mitigation persist. Instances like the AI-powered chatbot in New York City providing erroneous guidance underscore the need for robust safeguards. Additionally, there are concerns about potential job displacement, misinformation, privacy infringement, and automation bias associated with AI implementation.
While regulating AI in the private sector remains challenging, many state governments are exploring its application for public good. California's approach includes transparency measures, requiring companies to disclose the large language models used to develop AI tools, aiming to foster public trust.
Experts emphasize the importance of thorough testing and continuous monitoring of AI tools to mitigate risks post-deployment. Meredith Lee, from UC Berkeley's College of Computing, Data Science, and Society, underscores the necessity of ongoing evaluation to understand and address potential risks effectively.
California's Department of Transportation seeks GenAI tools to analyze traffic data and propose solutions for congestion reduction and road safety. Similarly, the Department of Tax and Fee Administration aims to leverage AI to enhance call center operations, reducing wait times and call duration.
Nick Maduros, director of the Department of Tax and Fee Administration, emphasizes that these tools are intended to assist rather than replace state workers. The envisioned AI technology would streamline tasks for call center workers by providing relevant tax code information during calls, potentially improving service efficiency.
Although deployment timelines remain unspecified, Tong asserts the state's commitment to swift progress, emphasizing the agility of GenAI technologies in implementation.
The initiative reflects California's proactive stance in harnessing innovative technologies to address public service challenges, with GenAI poised to play a pivotal role in enhancing government operations while navigating potential risks responsibly.