“That can be easy to do if it’s 50 rows of data. But when we’re talking about 10,000, 15,000, 20,000 lines of Excel data that needs to be understood, it’s really hard to try to do that,” said Liang, a data analyst with the San Jose 311 team.
Liang created a custom AI assistant that could take those thousands of pieces of data and arrange them into themes and categories, allowing trends and other pieces of information to emerge that can then be shared with various departments.
“I was looking for a way to make that process a little more efficient to not only improve my work, but also be able to quickly understand what residents are reporting and catch the trends that might happen seasonally, or at different times of the months," Liang explained. "And that’s exactly what this AI assistant was able to help me do."
Liang is not a deep expert in developing AI tools. He picked up both his interest — and expertise — in AI from a 10-week city-led training initiative called the AI Upskilling Program. The program is a partnership with San Jose State University that aims to help city workers use AI responsibly and effectively.
The program is fundamentally about the city developing ways to reduce time spent on tasks, while also improving resident services, said Albert Gehami, San Jose’s chief AI and privacy officer. That means empowering frontline workers to develop AI skills and tools tailored to their tasks.
The city staff, Gehami said, are already “experts in their space” — whether that involves grant writing, addressing resident concerns or whatever is involved in the various departments the workers operate in.
“They know their problems better than anyone else. They know their work better than anyone else,” he said. “Very candidly, we’re helping them say 'hi' to the AI and showing them how to use it. And then they start coming up with all these brilliant ways around how it can streamline their work.”
The city is wrapping up its second 10-week cohort, which includes about 30 staffers. About 20 staffers, including Liang, participated in the first cohort. Participants dedicate one hour a week to the training and ultimately earn an AI Upskilling credential badge, which can be added to their skills and education on sites like LinkedIn. The third cohort is set to begin this fall.
The upskilling program includes two tracks: The AI Upskilling track is geared toward office workers and administrative staff, with the goal of improving efficiency about 10 percent to 20 percent; the Data Upskilling track is designed more for data analysts, with training in tools like Power BI, Tableau, SQL, Python and GIS.
The goal is to offer the upskilling training to about 1,000 of San Jose’s 7,000 city workers next year. The city also offers more introductory AI training intended for a wider cross-section of workers.
“We have provided that to hundreds of staff members across the city,” said Gehami.
The city would like to continue to expand its introductory course to share with at least 15 percent of its staff and then also develop a few AI experts from the upskilling program within each division across the city.
San Jose has been a leader in understanding how AI can be effective in delivering public-sector services and developing the needed regulatory guardrails around this emerging technology cluster. The city is a founding member of the GovAI Coalition, formed in 2023 and which now includes more than 1,700 professionals from about 550 government organizations.
“One of the fundamental beliefs across both how we operate in the city, in IT, and also in the GovAI Coalition is that from the public perspective we are one government. We are all here to serve our communities,” Gehami said. “And the coalition has provided us so much value in staying on top of the technology, and being able to share these types of resources has been essential.”
The AI Upskilling course is public and has been shared with the coalition.
“And that’s benefiting not just San Jose, but is benefiting countless other communities and cities and states,” said Gehami.
The Oregon Department of Enterprise Information Services (EIS) also recently launched a program to train state workers around the responsible use of generative AI.
The free training will be optional, with two courses offered, in partnership with InnovateUS. One course, called “Responsible AI for Public Professionals," is two hours of self-paced instruction covering the practical use of GenAI in day-to-day work, as well as ethical and responsible AI practices. The course will also explore improving public service delivery with AI, said Hope Hiebert, communications director with EIS.
Another course, “Responsible AI for Public Organizations,” is also two hours of self-paced instruction, and covers the different types of AI and addresses ethical considerations.
“As is the case with all technology, it is a responsible thing to do to give everyone the opportunity to learn the emerging technology,” Hiebert said in an email. “The state of Oregon plans to continue to manage this in a meaningful, responsible, governed way.”
Back in San Jose, tech leaders are hopeful that the careful use of AI can help to save time on tedious tasks while improving productivity.
“We wanted to ensure that we’re not just maintaining the same level of performance, but wanted to be able to show our residents that we’re getting better year-over-year,” said Stephen Caines, chief innovation officer for San Jose.
Already, officials estimate the two cohorts that have completed the AI Upskilling Program have saved the city $50,000 and 10,000 to 20,000 staff hours, largely a result of reducing consulting costs related to AI development and maintenance.
“It just makes sense for the capital of Silicon Valley to improve quality of life for our residents by embracing innovation," said San Jose Mayor Matt Mahan in a statement. “Through our AI Upskilling Program we’re empowering our workforce with real-world AI skills that drive efficiency, reduce costs and improve the delivery of public services."
Liang, who developed the AI assistant to help analyze 311 data, has seen what happens when he’s not spending countless hours sorting through resident questions and concerns: Those same questions and concerns can get addressed faster.
“We are responding a lot more efficiently, and better understanding our residents,” Liang said, as he explained how AI helps the city to analyze certain requests for services within particular date ranges and get a sense of the top themes residents are reporting.
“And through that, we’re able to take action on those steps internally, whether that be to release more FAQs with questions that people are not getting answered, or able to find on our website, or just letting our agents know more people are contacting about this, and here’s the information you need to provide to residents,” he added.
