The Scene
San Jose Mayor Matt Mahan was on a bus recently to witness how AI was speeding up the notoriously slow mode of transport. It turned out the new system worked too well, with buses running ahead of schedule.
For Mahan, an entrepreneur-turned-politician in the heart of Silicon Valley, a too-efficient bus system was a welcome one, he joked, and is only one of the many problems he hopes AI can help solve.
“We should be able to automate much more of what we do in government, and that’s the journey that we’re on now,” he said in an interview with Semafor.
Though less glamorous than San Francisco’s startup scene, San Jose may be the ideal test case for government AI. The city is home to Nvidia’s headquarters and a cluster of semiconductor companies crucial to AI development. With 1 million residents, it’s 20% larger than San Francisco by population, and San Jose State produces more engineers than Stanford and Berkeley combined, reflecting 8% of California’s total graduates in the field each year.
It is also located in a state that is facing a budget shortfall of $20 billion out of a total of just over $200 billion, with long-term fiscal problems and politics that stand in the way of solutions. Software, California’s technologists often advocate, could be one answer.
Leading the charge in San Jose is Mahan, who co-founded a startup with Sean Parker of Napster and Facebook fame, and who gets advice from some of Silicon Valley’s biggest names. Under Mahan’s tenure, the city’s vehicles have been outfitted with pothole-detecting cameras and its employees use AI to win multimillion-dollar grants. He is even mulling ways that technology’s predictive power could be used to solve the seemingly intractable homelessness problem, by using automation to identify those at risk of ending up on the streets and trying to preempt them from hitting bottom.
Mahan helped start the GovAI coalition, convening local government leaders from around the country in San Jose to share best practices on how to use AI in government. The group shares assessments on new AI tools, serving as a kind of Amazon review section for tech-forward mayors around the country.
“What I kept hearing from people was that they just want government to work,” Mahan said. “The public sector just hasn’t gotten more productive or efficient. In fact, arguably, it’s gotten worse.”
In this article:
Know More
One of the downsides of the software era is that cities and governments have ended up with dozens, sometimes hundreds of “software-as-a-service” solutions, many of which are only minimally effective.
In San Jose, the cost of these solutions adds up to tens of millions of dollars or more, Mahan said. “We spend a lot of money just training people on how to use the tools that probably have way more features than they need because they’re trying to serve too many different customers and use cases,” he said.
For instance, San Jose pays for grant-writing software, but a city employee recently built a custom chatbot and was able to secure a $2.5 million grant after using it with minimal effort.
Mahan sees opportunities for automation anywhere there is inefficiency, such as the building permit process. “The permitting process is maddening, and part of what’s wrong with it is it actually allows too much human discretion,” he says. When there are clear guidelines and rules about what can be built and where, he said, there’s no reason the process should drag on for years.
“It’s shocking to me at the end of the day, it’ll take two years to get all the permits you need to build the very thing that we’ve already zoned for and said we want.”
To accelerate more AI adoption, Mahan helped implement an AI upskilling program, a 10-week training scheme run in conjunction with San Jose State that teaches city employees to leverage AI tools in creative ways.
Mahan said the program, after its first cohort, has already saved the city more than 10,000 hours of employee time.
The city also handed out grants to AI startups in an effort to attract young tech founders to San Jose when they might have otherwise gone to San Francisco.
“We’re trying to pitch them on the city being their first client,” Mahan said.
Reed’s view
Governments, especially in California, are going to need to automate their processes to cut costs. It won’t be as easy as they might hope, though. As the private sector is finding out, it takes a lot of work and infrastructure to get the most out of new AI capabilities.
Just training employees on AI tools will only go so far. There aren’t really off-the-shelf solutions that can move the needle. But the latest AI capabilities could allow governments to leverage software automation in ways that weren’t possible or feasible before. What governments need to do is transform themselves into builders of software, which probably means hiring talented engineers to build and implement AI.
Notable
- San Jose’s experiment of using AI to identify homeless encampments — the first of its kind in the US — has been criticized by local outreach workers, who worry it might be used to punish people at risk, The Guardian reported last year.
- President Donald Trump’s AI Action plan, launched in July, may support AI literacy and workforce development to prepare employees for how the technology will change their jobs, Council on Foreign Relations experts argue.