Here in the capital of the AI boom, San Francisco, the city government has just rolled out a ChatGPT-powered tool for use by city workers — hopefully they'll still do their own fact-checking!

Mayor Daniel Lurie announced that, starting Monday, Microsoft 365 Copilot is being made available to most of the city's 30,000 employees for helping with various "tasks including writing reports, data analysis and document summaries." Per the Chronicle, the software is powered by OpenAI’s GPT-4o chatbot, and the city has laid out some guidelines for its use.

"Avoid public or consumer AI tools unless formally vetted — never enter sensitive or City data into them," says one guideline. And another, "Always check the output. AI isn’t always right. Review, edit, fact-check, and test everything it generates."

What tasks will Copilot be most useful for? That will probably depend on the worker and the department. A review of the product by PCMAg in April suggests that it is handy for contextual research.

"San Francisco is the global home of AI, and now, we’re putting that innovation to work with  Microsoft Copilot Chat — allowing City Hall to better deliver for our residents,” Mayor Lurie said in a statement. "As our city and the world embrace AI technology, San Francisco is setting the  standard for how local government can responsibly do the same."

As novices to AI tools begin putting them to work across various walks of life, we can anticipate some snafus. Let's hope they don't end up as embarassments at Board of Supervisors meetings!

The city's guidelines prohibit the use of AI "to create City official documents or make decisions without expert human review."

And, classified under "medium- to high-risk" uses of generative AI in city government is "Drafting or translating public-facing content," and "Summarizing policy-related data." One can only imagine the potential embarassments there, if proper editing and fact-checking isn't always done.

Photo by Duo Nguyen