Before the mayor of San Jose, California, arrives for a ribbon-cutting event for a new business, his team asks ChatGPT to assist with drafting key talking points.
“Elected officials are often engaged in public speaking,” he explained, mentioning that his recent schedule has taken him from new restaurant openings to a festival celebrating low-rider automotive culture.
While some politicians might hesitate at the thought of a chatbot co-writing speeches and contributing to a $5.6 billion budget, Mahan is keen to adopt artificial intelligence technology. He aims to enhance the efficiency of nearly 7,000 employees in Silicon Valley’s largest cities by setting an example.
Mahan believes that embracing AI tools could streamline processes, allowing the city to serve its approximately 1 million residents more effectively.
He’s not alone; executives in both public and private sectors are pursuing aggressive AI strategies. Yet, there are concerns that expensive AI solutions can sometimes complicate things and lead to errors.
“The goal is to experiment, be transparent, and identify challenges,” Mahan shared in an interview. “We need to collaborate with different agencies and vendors to address issues together. New technology always has its hiccups.”
The city plans to train about 1,000 workers, or roughly 15% of its staff, on AI usage by next year. They’ll use these tools for various tasks, including addressing pothole complaints, optimizing bus routes, and employing vehicle surveillance to aid in crime solving.
One of the early adopters in San Jose was Andrea Arjona Amador, who leads the Electric Mobility Program within the transport department. Utilizing ChatGPT, she secured a $12 million grant for electric vehicle chargers.
Arjona Amador has developed a tailored “AI Agent” to help manage communications for grant applications and organize necessary incoming information. She played a key role in drafting a comprehensive 20-page document.
So far, San Jose has invested over $35,000 in 89 ChatGPT licenses for city employees, at $400 per account.
“We used to put in countless nights and weekends trying to finalize grants,” she remarked. Following a setback when the Trump administration withdrew federal funding, she successfully approached local backers with a similar proposal.
Arjona Amador, who has a background in Spanish and French before becoming proficient in English, has also created another custom chatbot to refine the tone and language of official communications.
Connected with major tech players like OpenAI and Google, the mayors of the Bay Area cities are eager to advocate for AI adoption that aligns with the industry’s aspirations, while also establishing guidelines to mitigate risks.
San Francisco’s Mayor Daniel Lurie announced a plan to give close to 30,000 city employees, including nurses and social workers, access to Microsoft’s Copilot Chatbot. The initiative emphasizes “robust privacy and bias protection measures and provides clear guidelines to reinforce technology, rather than replace human decision-making.”
San Jose has similar protocols and has yet to report any significant issues in its pilot projects. This attention to detail is crucial, especially given the rising concerns about misinformation that can come from AI-generated outputs.
For example, a flawed document released by the US Health Secretary Robert Kennedy Jr.’s “Make America Healthy” committee showcased the potential pitfalls of relying on AI. In another instance, Fresno school officials resigned after putting too much trust in an AI chatbot’s generated information.
While some government bodies may keep their use of chatbots under wraps, Mahan is transparent about the background notes he derives from the chatbot, which he frequently refers to during speeches.
“In the past, it would take hours of calls and reading, and you wouldn’t necessarily glean those insights,” he noted. “Now, you can accomplish these tasks at a similar, if not better, quality much more efficiently.”
However, he emphasized the importance of human oversight. “You can’t just push buttons and trust the output. Independent verification and logical reasoning are essential,” he cautioned.
Earlier this year, OpenAI unveiled a new pilot tool called Operator, intending to surpass standard chatbot functionalities by managing calendaring tasks and other responsibilities. The development of such “AI agents” is a priority for the tech industry right now.
If we look eastward from Silicon Valley, the Central Valley is also considering what these AI agents could accomplish. There’s potential for the Parks and Recreation Department to use AI to streamline park or swimming pool reservations, or for residents to check pool capacity before heading out.
Unfortunately, after completing an initial proof of concept, the city opted not to purchase a full technology license due to high costs.
Market research firm Gartner has forecasted that over 40% of “Agent AI” projects may be halted by 2027.
Still, the mayor of San Jose remains optimistic that these AI tools can indeed expedite the often cumbersome bureaucratic processes.
“Large organizations face an enormous amount of bureaucracy,” Mahan said. “Roles such as finance, accounting, HR, or grant writing could see a productivity boost of about 20%.”
