When it comes to artificial intelligence, San Francisco officials are looking to get ahead of it, rather than play catchup like they have with other technologies (taxi apps, electric scooters, and Airbnb all come to mind).
Earlier this week, the city’s Board of Supervisors approved outgoing District 9 Supervisor Hillary Ronen’s ordinance to require city departments to report their use of AI. The new rule has had a “surprisingly easy” time making its way through the city’s legislative channels, Ronen told Gazetteer SF.
“Part of me feels we’re at the very beginning of policymakers realizing what a big deal [AI] is and I think the fact that this just glided on through is an indication of that, which makes me even happier that I did it,” Ronen said.
The legislation will require SF’s chief information officer, Michael Makstman, to create a public inventory of AI use across city departments, along with each tool’s purpose, accuracy, biases, and limits. The rule also mandates that city department heads detail what data the AI is collecting and any potential risks associated with the technology. Makstman will have six months, beginning 30 days after Mayor London Breed signs the ordinance, to begin collecting and publishing the inventory on DataSF, the city’s open data platform.
Saira Hussain, a staff attorney with the Electronic Frontier Foundation, applauds the city’s efforts to be proactive, she told Gazetteer.
“There’s been an explosion around AI tech and a lot of people think it’s the silver bullet to do everything with,” Hussain told Gazetteer. The best way to vet uses of the new tools is telling the public “how the tech is being used, and for what purposes,” she said.
The goal of the legislation is to foster transparency, Ronen told Gazetteer. She’s been following the rapid development and expansion of AI products, and was surprised to learn that the city’s own department of technology “had no idea what AI the city was using,” in part because most of the city’s numerous departments have their own IT units, she said.
“There are opportunities and many risks in using AI products,” Ronen said. “I thought at the very least policy makers and the public should know what the city is using.”
Maktsman, who will be tasked with implementing the rule, told Gazetteer that he’s already aware of several city departments that have used AI, including the Department of Technology using AI to detect cybersecurity threats, the city’s 311 app using AI to suggest service remedies to issues based on someone’s photo or description, and the Department of Pubic Health’s radiology department using an AI-based medical imaging tool to confirm stroke diagnoses.
While those are all seemingly innocuous uses, Hussain cautioned that if governments begin to use AI technologies to “make certain determinations or decide who gets a certain service,” it could “have impacts on people's rights, could perpetuate bias and could engage in discriminatory outcomes.”
Hussain hopes the legislation will lead to protections for communities vulnerable to that kind of harm, such as legislation to require actual analysis of the technologies and their deficiencies, rather than a self-reported survey.
While Ronen’s tenure as city supervisor ends this year, to be succeeded by fellow progressive Jackie Fielder, she thinks her legislation will serve as a foundation for future regulation of AI. Ronen told Gazetteer she hopes city officials will develop explicit processes around procuring and using AI products, for transparency and to keep people safe.
“I am of the group of people who really believe [AI] is radically going to change our world,” she said. “I just want to make sure that voices outside of big tech are included in thinking through and influencing what technology gets developed, and for what purpose.”
Editor's note: This story was updated at 11:07 a.m. on Dec. 13, 2024 to correct Michael Maktsman's name. His name is Michael, not Matthew.
