If an elected representative is willing to utilize unregulated technology to do their job for them, that may raise an eyebrow; but if that official were to also suggest that the technology filter out any unnecessary conversation from government meetings, that should raise concern.

Ideally, Artificial Intelligence (AI) is a helpful tool which can be used to transcribe a meeting’s minutes, a job expected of an appointed clerk or secretary who themselves are oath-taking members of the very government board or commission whose minutes are being recorded. The reality of AI, however, is more dense and subversive than the lofty expectations of what it can achieve.

On June 22, 2023, the Nederland Sustainability Advisory Board (SAB) appointed Trustee Nichole Sterling, the Board of Trustees (BOT) liaison, to the role of Secretary, a role which Sterling had expressed she felt was almost unnecessary and that she would accept only if it was understood that she would be relegating the duties expected of her as Secretary to an AI program.

“Do we need a secretary appointed or can we just rely on someone hitting the record button?” Sterling asked as the SAB were looking for candidates to run for specific and important roles on the board. “Can I just be a secretary and utilize a robot?” she said.

Town Administrator Dr. Miranda Fisher stated during the meeting that the board is only legally required to include decisions on Action Items within the official minutes. SAB Chair Melody Baumhover noted that having discussion included in the minutes does help with Board operations, despite the fact that it is not a legal requirement.

“It’ll be heaven, you won’t have to do anything,” former Board member Larry Tasaday said. “AI is smart enough to figure this out and we almost won’t need a secretary anymore; how fabulous that would be!”

Sterling also expressed that the AI could filter out any unnecessary discussion in the SAB’s meetings.

“You want an AI program that sifts through all the non-important conversation and just bubbles up the action items,” she said.

The implications of what such “sifting” could eventually lead to in relation to allowing AI to assimilate government roles are great when considering how AI software is still being assessed under moral and philosophical examination. It is also beginning to be analyzed under a legal microscope.

Sterling’s idea to use AI in Nederland government came specifically from having worked with similar software in her career. Trint, otter, NOVA, and Speak AI are just a few of the most popular transcription software powered by AI, used in a variety of applications in the business world. Generative AI, which can be prompted to “create” documents, imagery, audio, and synthetic data, is most used by businesses for marketing and sales, operations, and risk and legal.

However, as businesses leap into using such programs to easily record their meetings for accuracy and permanence, the free use of such unregulated technology can lead them to legal troubles down the road if litigators were to ever request copies of any automatically generated or AI-scripted transcripts or summaries of any meeting during which a particular subject under scrutiny was discussed.

During such a scenario, how does AI fall under the rules of evidence? According to several US courts, “machine testimony,” such as computer-generated test results or speed monitoring devices, is not hearsay and is admissible as long as the record is “the result of a process free of human intervention.”

Under this provision, would evidence from AI be considered machine testimony, and therefore would the evidence be admissible considering that AI is, in most uses, dictated by prompts from a human being? The regulatory language needed to explain the answer to this hypothetical question has yet to be written on either a state or federal level.

There have also been cases concerning whether AI-based videoconferencing violates state laws regarding two-party consent laws for recording private conversations without the consent of participants.

So far in 2023 AI has made several appearances in court cases across the country, from a New York lawyer attempting to pass by the judge AI-generated case research, to an AI smartphone app being used as legal aid defending a client from charges of a speeding violation.

The AI legal aid app, created by DoNotPay, was designed to listen to court arguments and formulate responses and relay them to defendants in real-time through headphones. The app is primarily used to handle customer service disputes and was to be implemented for the first time in US courts on February 22, 2023.

However, due to the technology itself being illegal in many courtrooms, DoNotPay’s CEO Joshua Browder was threatened by state bar prosecutors with the possibility of up to six months of jail time if he were to attempt to have his program represent a defendant in court.

Lawyers for the firm Levidow, Levidow, & Oberman, who had ChatGPT formulate their legal research in a personal injury lawsuit involving an airline, were disciplined in June of this year for using the program. The AI generated six legal cases that did not exist, to be used for reference in the case.

“Six of the submitted cases appear to be bogus judicial decisions with bogus quotes and bogus internal citations,” the judge stated in his order of explanation to the legal team.

One of the lawyers, after expressing regret for his decision to use ChatGPT, said that he was “unaware that its content could be false.” The lawyers gave testimony that they prompted ChatGPT to “doublecheck” that certain cases were real, to which the program responded that the cases could be found on several legal reference databases.

Texas Federal Judge Brantley Starr declared in May of this year that no court filings drafted by generative AI will be accepted in their court unless they’ve been adequately checked for accuracy.

“All attorneys appearing before the Court must file on the docket a certificate attesting either that no portion of the filing was drafted by generative artificial intelligence or that any language drafted by generative artificial intelligence was checked for accuracy, using print reporters or traditional legal databases, by a human being,” Judge Starr stated.

“These platforms in their current states are prone to hallucinations and bias. While attorneys swear an oath to set aside their personal prejudices, biases, and beliefs to faithfully uphold the law and represent their clients, generative artificial intelligence is the product of programming devised by humans who did not have to swear such an oath,” the judge continued.

“As such, these systems hold no allegiance to any client, the rule of law, or the laws and Constitution of the United States (or, as addressed above, the truth). Unbound by any sense of duty, honor, or justice, such programs act according to computer code rather than conviction, based on programming rather than principle.”

A Secretary, as a member of a board, would also be required to give an oath when being sworn to the role, as would a Town Clerk, whose official duties include drafting the minutes for the BOT.

Would a Board Secretary or a Town Clerk be in violation of their oath by willfully allowing an AI program to take control of part of their duties? Again, there is no language written to provide an answer.

This isn’t meant to be a condemnation of AI or of anyone who would recommend its potential as a useful tool. According to an MIT technological review, the implementation of generative AI in government, citing examples in Asia, can be a supportive tool to help public workers expedite the delivery of services to citizens, including in areas of public health and education.

Microsoft’s Regional Director of Government Affairs and Public Policy in Asia describes the implementation of such technology as less about governments saving money and replacing jobs in the public sector and more about “freeing up time for the more value-added human aspects of what it means to work in government.”

Microsoft Azure OpenAI is designed to assist in the menial task of searching for sorted and archived information, which is a heavily required task in government work. The claim is that with such work being so quickly streamlined to the user through the use of AI, it would free the government employee to better serve and engage the public.

Doctors would also have fast access to patient records, and teachers could presumably better prepare lesson plans and adjust to best assist a student’s particular needs in learning.

With such AI programs being pitched as the perfect “co-pilots” in government planning that create high-impact results at low costs, perhaps the SAB is the perfect testing ground for such a revolutionary experiment.

The Board’s focus is primarily on implementing green energy solutions for the town of Nederland through proposing actionable plans and seeking grant opportunities to facilitate recycling and solar energy programs, and other projects aimed at reaching the town’s 100% sustainability goal by 2025.

It is true that the SAB’s goals do not fall under public scrutiny as often as the BOT’s or Nederland Downtown District Authority’s decisions do. But would the implications and legal uncertainty associated with AI be acceptable if it were used to perform Secretary or Town Clerk duties on a BOT or Planning Commission meeting?

The MIT technological review noted that “it is vital the public sector develops and deploys generative AI responsibly, in ways that protect privacy and data security and fosters citizen trust.”

Referring to AI as a “tool” implies that it is operated by a person. However, the idea of filtering dictated conversation implies that the AI is being allowed to act on its own, although with set parameters.

Allowing AI to be used as more than a tool but as a resource, especially at this uncertain stage in its development and legality, should be a discussion treated with concern, and not accepted just on the grounds that it makes someone’s job easier, especially when it’s a job with the potential of coming under public scrutiny.