The very first order of business in Newsom’s Executive Order N-12-23, signed Sept. 6, was for the California Government Operations Agency (GovOps), the Department of Technology, the Office of Data and Innovation, and the Governor's Office of Business and Economic Development to partner with other state entities in drafting a report to the governor “examining the most significant, potentially beneficial use cases for deployment of GenAI tools by the state” — as well as potential risks. This is that report, from GovOps, and it offers six “beneficial use cases for GenAI in state government.”
In a statement, Newsom called the report “an important first step in our efforts to fully understand the scope of GenAI and the state’s role in deploying it.” He added: “We’re taking a nuanced, measured approach — understanding the risks this transformative technology poses while examining how to leverage its benefits.”
“Through careful use and well-designed trials, we will learn how to deploy this technology effectively to make the work of government employees easier and improve services we provide to the people of California,” said GovOps Secretary Amy Tong, who led the team behind the report, in the statement.
The use cases are:
- Instances that improve the performance, capacity and efficiency of the state’s ongoing work, research and analysis via summarization and classification. GenAI can analyze hundreds of millions of data points at the same time, creating comprehensive summaries of any collection of artifacts, irrespective of whether the content is in a text, audio or video format. As the technology learns, it could also categorize and classify information by topic, format, tone or theme. Specific use cases here include sentiment analysis of public feedback on state policies, with GenAI recommending change to improve process and service delivery; and summarizing meetings, work and public outreach documentation, using GenAI to identify insights, find the key topics, conclusions and action items.
- Instances that may personalize and customize “work products” across a diverse population of residents, with the potential for improving outcomes and access to service. Here, GenAI can recommend ways to display complex information in a way that resonates best with various audiences, per the report, or highlight information that may be relevant to an individual. Example use cases here are applying GenAI to government service data to identify groups or subsets that could benefit from additional outreach, support services and resources based on circumstances and needs, and identifying groups that because of “language or other reasons, are disproportionately not accessing services.” Here, the technology could analyze feedback, surveys or comments to identify language that indicates accessibility difficulties.
- Instances that improve language and the access to communication across multiple languages and formats. Here, with a single prompt, a GenAI solution could build a video or image for further user refinement — enabling the state to make videos, recordings and documents more accessible and inclusive. Use cases include using GenAI to help convert educational materials into formats such as audiobooks, large-print text and Braille; and using GenAI to help translate government websites, public documents, policies, forms and other materials into a variety of languages, thereby expanding access to people who aren’t native English speakers.
- Instances that optimize software coding and explain and categorize unfamiliar code. GenAI is capable of generating code in multiple computing languages and translating code from one language to another. This, the report said, can help state systems that use code written in obsolete languages and can facilitate state understanding of the precise technical architecture of agency applications. Use cases include code conversion tools based on foundation models that can accurately translate legacy codebases like COBOL mainframe apps into modern programming languages and GenAI development tools that can auto-generate quality code and create test environments and synthetic data sets to train machine learning models.
- Instances that find insights and predict key outcomes in complex data sets for decision-makers, enabling staffers with a range of expertise to reach findings that may be relevant to their work. Use cases include cyber protection systems powered by foundation models that can rapidly analyze network activity logs, identify anomalies and threats, explain attacks and suggest remediation, as well as analyzing data streams from drones, satellites and sensors to create detailed damage and deterioration assessments, improving maintenance forecasting.
- Instances that optimize workloads for environmental sustainability by optimizing resource allocation, maximizing energy efficiency and demand flexibility and promoting eco-friendly policies. Potential use cases here include analyzing traffic patterns, ride requests and vehicle telemetry data to enhance routing and scheduling for government-managed transportation fleets — potentially reducing fuel use and emissions — and, via GenAI simulation tools, modeling the carbon footprint, water usage and environmental impacts of large infrastructure projects. This could identify the most sustainable options.
