Artificial Intelligence At The County Government Center? Board Of Commissioners To Consider Policy

Leelanau County could soon have documented rules in place regarding the use of artificial intelligence (AI) for county workers. As part of today’s executive board session, county commissioners will consider a policy that limits when, where, and how employees can utilize AI in the course of their day-to-day work. The new policy comes at a moment when AI is becoming increasingly commonplace – and controversial.

Released in demo form on November 30, 2022, the AI-powered chatbot known as ChatGPT rapidly popularized the use of generative AI for all manner of everyday tasks, including research, emails, writing and editing, and even computer programming. ChatGPT now ranks as the fifth most-visited website in the world. It’s also one of the most controversial: OpenAI, the company behind the chatbot, is facing lawsuits for everything from copyright infringement to mental breakdowns and suicides.

Leelanau County’s draft AI policy does not call for an outright ban on the use of generative AI. In a memo to commissioners, County Administrator Jim Dyer wrote that the use of AI in the workplace “could be a useful tool to increase efficiency in some regular tasks performed by departments,” citing how the technology “quickly creates templates, outlines topics, and presents material in an easy-to-read format.”

“It is becoming clear that an employee may attempt to use this new technology without clear guidelines for its limitations, and the attached policy seeks to clarify some of those concerns,” Dyer continued. “The policy…was drafted by our IT director, edited by human resources, and approved by legal counsel.”

But the proposed policy does call for several major limitations. The policy would bar employees “from entering confidential information into the AI software.” It also requires employees using generative AI “to always review results for accuracy and completeness.” Perhaps most notably, the policy would limit the use of AI “for research and development purposes” only, noting that AI is “not a substitute for human judgment and creativity” and that “Al results should not be used as a final work product and should not be relied upon as the sole means for making decisions.”

The policy would apply “to any third-party or publicly available GenAl tools, including ChatGPT, Google Bard, DALL-E, Midjourney, and other similar applications that mimic human intelligence to generate answers, work product, or perform certain tasks.”

The draft policy focuses heavily on the protection of sensitive information. County employees would be prohibited from uploading or sharing “any confidential, proprietary, or protected data” with generative AI tools. “Once such information is disclosed, it is impossible to ‘undo’ the damage,” the policy states. This rule applies to passwords, credentials, protected health information, personnel material, “information from documents marked confidential, sensitive, or proprietary,” and “other nonpublic information that might be harmful if disclosed.” Employees would also be barred from inputting personal information – about themselves or anyone else – into a generative AI tool, including names, addresses, or likenesses.

Despite leaving the door open for most county employees to use generative AI, the draft policy does call out one department where the technology would effectively be forbidden: human resources.

“Do not use GenAl tools to make or help you make employment decisions about applicants or employees, including recruitment, hiring, retention, promotions, transfers, performance monitoring, discipline, demotion, or terminations,” the policy states.

Employees who do choose to use generative AI in the course of their work with the county would be expected to do so with extreme discretion. For example, such individuals would be required to inform their supervisors whenever they’ve “used a GenAI tool to help perform a task.” The policy also puts the onus on county workers to evaluate any AI tools they choose to use, including the content those tools produce.

“The use of Al results can produce out of date, copywritten, or misinformation,” one part of the policy cautions. “Employees should evaluate the utility, security, and reputation of the Al tool before using it. Employees should not blindly trust any output from these models and should always verify the facts and content they generate. Verify that any response from an GenAl tool that you intend to rely on or use is accurate, appropriate, not biased, not a violation of any other individual or entity's intellectual property of privacy, and consistent with the County's policies and applicable laws.”

If commissioners approve the policy as written, they’ll also be granting the IT department permission to “verify compliance…through various methods, including but not limited to network monitoring, business tool reports, internal and external audits, and inspection.” Any employee caught violating the policy “may be subject to disciplinary action, up to and including termination of employment.”

This executive board session is scheduled for 9:30am in the commissioner meeting room at the Leelanau County Government Center. The full proposed AI policy can be reviewed on pages 75-79 here.