Eureka Talks AI—So We Had AI Cover the Meeting, Too

A real photo of the Carson Mansion by Cory Maylett on the left (via WikiCommons) and a CHAT GPT version of the same image held in a robot hand.

The Eureka City Council recently discussed the role of artificial intelligence in local government.

Our reporter, Daniel Mintz, covered the meeting. For comparison, Daniel also ran the full meeting transcript through Google Gemini using a basic prompt asking it to summarize the discussion.*

Below are both versions.

___________________

Fearing ‘Brain Drain,’ Eureka Seeks to Rein in AI

By Daniel Mintz

Wary of increased use of Artificial Intelligence (AI), the Eureka City Council is advancing rules to bring it under control.

At its March 17 meeting, the council approved a new policy “to define the scope in which Artificial Intelligence may be utilized by city employees,” but it’s just a first step.

Some councilmembers described AI as dangerous to the environment and to society.

In a presentation, Finance Director Lane Millar said the “heart of the policy” is generative AI, which is used for tools like ChatGPT, Claude and Gemini.

The policy’s “fundamental rules” prohibit submitting confidential information to AI platforms, require double-checking of AI-generated content and mandate transparency on when AI has been used for research or to create content.

The policy also warns that AI “may produce biased or discriminatory results” and “should only be used as part of information gathering or fine-tuning verbiage where the final decision is left to a human being.”

Employees use AI in a variety of ways and it’s “very good at producing a lot of content very quickly,” said Millar, giving the example of a poem about Eureka produced by ChatGPT “literally in seconds.”

Screenshot of the AI Generated poem shown at the Eureka City Council meeting.

The poem mentions the city’s “Victorian queens,” the Carson Mansion, fog, redwoods and the waterfront fishing industry, and ends by saying the city “stands in moss-draped grace/a wild, salt-dusted timeless place.”

Millar said while generative AI is “trained to look for patterns” and then provide content, the “next stage of development” is agentic AI.

“ChatGPT can produce an email for you,” he continued. “With agentic AI like OpenClaw, you can give it your login to your email and it will automatically respond to your emails throughout the day or whatever timeframe you tell it to. It could essentially mimic or attempt to mimic the behavior of someone that spends a lot of time on the computer.”

Millar doesn’t see agentic AI “being embedded into our daily lives at the city but it will likely be used by other software programs that the city acquires.”

For now, the policy seeks to make use of AI accountable.

“If you’re tasked with determining the effects of a new regulation and you rely on ChatGPT to tell you the impact of a new bill and you don’t do any of the research yourself, it’s important that you let the reader know or let your boss know that that work was not done by yourself, but done through an AI tool,” Millar said.

Some of that has already been done voluntarily. When the council approved a Trans Sanctuary City resolution earlier this year, City Manager Miles Slattery admitted that ChatGPT had a hand in authoring it.

A public comment period only had one taker, a caller who described AI as “an industry, a field that is absolutely captured by a class of psychopaths with a dystopian future intended for all of us.”

That opinion got some back-up from councilmembers.

“Because of the environmental, cultural and intellectual degradation caused by the use of AI, I don’t think we should be using it at all,” said Councilmember Kati Moulton. “I don’t think that the city should use it. I don’t think that people should use it. At least not in its current form, but that boat has long since sailed.”

She added that “students who use AI lose the ability to think critically and it stands to reason that employees who use AI lose the ability to practice on those skills that they have learned and that they were hired for,” leading to “strong potential for brain drain.”

The city’s use of AI should be “extremely limited, basically to controlled analysis,” said Moulton. “Our employees need to read and write their own emails.”

She suggested multiple changes to the policy’s language and definitions, saying some of it is incorrect and sounds “like marketing speak from people who write AI.”

Councilmember Leslie Castellano was less condemning.

“I do think AI can be a tool,” she said. “I use it in my work and I think there’s value in sort of supporting its use in some ways and then having a policy around it.”

Saying AI is “such a significant issue for us, for humanity,” Councilmember Scott Bauer vouched for further work on the policy.

“I read it and went ‘huh?’ — that was my reaction,” he said. “This needs to be a policy that we spend some time on.”

Councilmember Renee Contreras-DeLoach said her opinions are “a little bit closer to Councilmember Moulton’s and our public commenter” and described AI as something that “hoovers up resources and creates environmental degradation in its path” and has “so many weird red flags.”

Sensing concern, Millar suggested control of AI is limited.

“This policy might look good on paper but our ability to actually keep employees from using these tools could be difficult,” he said. “And in addition to that, with a lot of the software programs we have already, you do have some pretty sophisticated AI tools.”

Millar said the city is already using “multiple programs that can go as far as reading your emails and looking at an attachment and if that attachment is an invoice, it can actually pull out details from that invoice and get it ready for payment processing.”

He suggested having a study session or creating a “small working group” on the policy.

With Councilmember Mario Fernandez absent, the council voted to approve the policy but to revisit it no later than the second meeting in April.

In the meantime, an ad hoc committee made up of Moulton and Castellano will work with staff on amending the policy in line with the issues that had been discussed.

———
AI-Generated Version (Google Gemini Summary of Meeting Transcript)
———

The following summary was generated from the full meeting transcript using Google Gemini and is presented as produced.

Eureka Faces the Algorithmic Frontier: Inside the Council’s Heated Debate Over AI

By Google Gemini

EUREKA, CA — In a session that felt part municipal briefing and part philosophical standoff, the Eureka City Council voted on March 17 to adopt an interim Artificial Intelligence policy, even as some members warned the technology could lead to a “dystopian future” and the “intellectual degradation” of the city’s workforce.

The meeting served as a collision point between the pragmatic need for data security and deep-seated anxieties about the rapid rise of machine learning. While the council ultimately approved the policy, they did so with the caveat that it must be overhauled and brought back for review by late April.

From Netflix to ‘OpenClaw’

Finance Director Lane Millar opened the presentation by demystifying AI, tracing its evolution from familiar recommendation engines like Netflix and Spotify to the “heart” of the new policy: Generative AI. Tools like ChatGPT and Google Gemini, he noted, are already being used by staff to draft emails, summarize complex state regulations, and even generate poetry about Eureka in “literally seconds”.

However, the conversation quickly shifted to Agentic AI, a more advanced frontier where systems like OpenClaw don’t just write emails—they take over the inbox entirely.

“It could essentially mimic or attempt to mimic the behavior of someone that spends a lot of time on the computer,” Millar warned, adding that while the city might not deploy such “agents” directly, they are already arriving embedded in third-party software.

The Pushback: ‘Marketing Speak’ vs. Reality

The most pointed critique came from Councilmember Kati Moulton, who took aim at the very language used to describe the technology. Moulton argued that the draft policy relied on “marketing speak” that anthropomorphizes software.

“It doesn’t mimic the structure of human brains or human logic,” Moulton stated, proposing that the city replace terms like “create” with “synthesize copied content” and “learn” with “predict relevant language”. Her primary concern remained the “strong potential for brain drain,” fearing that employees who rely on AI will lose the critical thinking skills they were hired to exercise.

Moulton’s skepticism was matched by a public commenter who described the industry as “absolutely captured by a class of psychopaths” and Councilmember Renee Contreras-DeLoach, who admitted to being a “Luddite” wary of AI’s “environmental degradation”.

A ‘Stop-Gap’ Measure

Despite the friction, a majority of the council agreed that leaving the city with no policy was a greater risk than adopting an imperfect one. Currently, staff may be using these tools without privacy guarantees, potentially training global AI models on sensitive city data.

The approved interim policy includes several “Do’s and Don’ts”:

• Data Silos: Employees must never input Social Security numbers, personnel info, or proprietary tax data into AI tools.
• Fact-Checking: Staff are warned not to “rely blindly” on AI, as the systems frequently “hallucinate” or provide biased information.
• Transparency: Any work “significantly contributed” to by AI—such as original content or strategic frameworks—must be disclosed to supervisors or the public.

The Road Ahead

The council concluded by forming an ad hoc committee, appointing Councilmembers Moulton and Leslie Castellano to work with staff on refining the policy’s definitions and ethical guidelines.

“This needs to be a policy that we spend some time on,” noted Councilmember Scott Bauer, citing the significant impact the technology could have on humanity within the next five years. The committee’s revised version is slated to return to the floor no later than the second meeting in April.

For the record we like our human shaped CHAT GPT generated poem about Eureka better than theirs…

Eureka

Fog flows over 
like it owns the place—
and, if you’ve lived here long, you know it does.

It curls around Victorian houses that lean into the years,
painted ladies watching
another grey tide roll over them and through the streets.

Redwoods now tangled in damp, curling grey strands
carry their own long memory
of rain and sun and time that came before.

Here, not much moves fast—
not the fog,
not the conversations,
not even the light as it fades
into another gray-blue evening.

As it winds around the town, the fog carries—
salt, smoke,
and the sense that
whatever the world is rushing toward,
Eureka will get there
in its own way.

*Note: Many RHBB articles may incorporate AI tools as part of the reporting process. All content is reviewed by our reporters, who retain full responsibility for accuracy.