The East Bay is a test case for cities turning to AI tools
Most of us haven’t noticed, but our East Bay local governments are in the middle of a revolution in how they deliver services to the public. City halls and other agencies are quietly but quickly incorporating Artificial Intelligence (AI) tools into their workflow.
This raises some basic questions. How is AI performing? What issues have come up? What’s the total cost?
So I set out to uncover which East Bay public agencies are using AI, and what they’re doing. Even with AI systems assisting my search, answers are scarce. Transparency seems to have fallen behind the rush to modernize. A tech executive once made the news when he tried to persuade President Trump to declare a national emergency in order to build a 512-acre deregulated AI research zone on the island. Nothing happened. Yet actual AI adoption is now unfolding all around us with far less attention.
Part of the challenge is the sheer number of overlapping public agencies that govern life on Alameda Island. We are part of the City of Alameda, the County of Alameda, and several independent boards and districts. Each one could be experimenting with AI, and each one could be signing contracts that the public never sees.
The County of Alameda is the biggest agency here, and that’s where the worries begin. Things looked promising last year when the County published what appeared to be robust guidelines that proposed limited use of the tool and careful balance of its risks. Obviously, local government desperately needs more efficiency, so AI sounds like a dream. The County kicked off with high-profile programs like a chatbot to help access documents and data tools to streamline the budgeting process.
One year later, it is impossible to know how these tools are performing. There is no public dashboard of successes, errors, or costs. My own AI search tool turned up nothing, and my Public Records Act request is still pending.
This lack of transparency is a red flag. Should we just assume that the chatbot never hallucinated, that privacy was never violated, and that humans kept close control over sensitive issues? Probably not. Researchers at UC Berkeley simply warned us not to believe the “myth of universal productivity enhancement” by AI. Companies frequently drop their AI projects when they have problems, but that isn’t really an option in the public sector.
Other East Bay agencies left a few more bread crumbs. AC Transit recently procured an AI system to automatically generate parking tickets. The system quickly and mistakenly sent out tickets to parents doing school drop-offs. It is unclear whether officers actually reviewed these tickets before they went out, as was promised. This is a serious issue, because in California people have already been kept in jail because of evidence that was AI slop. The widespread adoption of automated license-plate recognition (ALPR) across the region could make this even more worrisome if mistakes are made. Errors have real consequences.
Some Alameda agencies provide partial clarity. The Alameda Police Department posts an AI use policy, although it does not report outcomes or risks. I could not find any AI policy from the City of Alameda or the Alameda Unified School District, even though we know that individual employees are certainly using the technology to do their work.
This unclear and uneven rollout is just one example of the national trend of cities adopting AI tools. Bloomberg Philanthropies found that 69% of mayors surveyed are testing the technology. The National Association of Counties found that 60% of county officials who responded already use generative AI. It’s also happening in the midst of a budget crunch hitting local governments everywhere. Public agencies that are under financial strain may be tempted to cut corners, adopt systems without strong oversight, or ignore red flags.
So who is putting the AI into Alameda Island, and how well is it working? Is it worth the investment? Right now, the public doesn’t really know. Our early experiences here in the East Bay show that we should ask tougher questions and demand better answers about how AI is being used locally. It’s better to set up guardrails in advance than to respond to emergencies afterward.
Shum Preston is a Certified Information Privacy Professional (CIPP-US) from Alameda. He was previously Director of Communications for the California Department of Justice and Director of Communications for Common Sense Media.
Editorials and Letters to the Editor
All opinions expressed on this page are the author's alone and do not reflect those of the Alameda Post, nor does our organization endorse any views the author may present. Our objective as an independent news source is to fully reflect our community's varied opinions without giving preference to a particular viewpoint.
If you disagree with an opinion that we have published, please submit a rebuttal or differing opinion in a letter to the Editor for publication. Review our policies page for more information.
