By Jake Dye

Special to the Homer Independent Press

Nearly two years after Kenai Peninsula Borough School District Superintendent Clayton Holland said an internal committee would draft guidelines for the use of generative artificial intelligence tools in local schools, the district’s Board of Education on Monday got its first look at a set of proposed policies.

The policies add AI chatbots and other websites that utilize AI tools to a list of media regulated by the district for communication between staff, students and parents. They say that inputting student data into a chatbot or AI tool would constitute revealing that data to a third party in violation of student records policy. They say that AI use “must adhere to the principles of academic honesty.”

“Some AI tools are recognized as an extension of learning resources, supplementing but not replacing student originality and effort,” one policy reads. “Students are expected to use AI tools as an enhancement to their learning, not as a substitute for their intellectual work.”

It was the latter policy that drove most of the conversation Monday. Mica Van Buskirk called “unacceptable” language in the proposed academic honesty policy that says AI may be used for “development of original thought and material.” She also pushed back on the characterization of the technology in the policy more broadly.

“I have pretty deep concerns about the way this is worded, because I feel like it implies that AI is broadly accepted and encouraged and it doesn’t give any clear distinction between what’s acceptable and what’s not acceptable,” she said. “AI doesn’t develop original thought and material, it pulls in others’ original thoughts and material.”

The draft policies were presented during a meeting of the policy review committee ahead of a full meeting of the board this week. They’ll be considered further at a currently unscheduled work session in August. 

The policies were recommended by an internal AI committee, assistant superintendent Kari Dendurent said Monday, that included district professional development coordinator Amanda Adams, director of information services Eric Soderquist and school board member Ash-Lee Waddell.

Generative artificial intelligence, as it’s come to be called, largely refers to a suite of technologies built on large language models that create outputs from massive datasets in response to prompts from users. Those outputs can come in many forms, including text, documents, images and sometimes videos. 

The responses are based on and extrapolated from content that the models have been trained on, though often without citation, and many models have been documented to sometimes “hallucinate” or provide inaccurate or entirely fabricated responses. 

Generative AI tools have been integrated into many familiar programs and websites, including social media, search engines and other software like email or word processors. Members of the committee said Monday that it isn’t always clear to users when they have begun to use the technology. Some users may be unaware of potential shortcomings or pitfalls.

In one instance, the Alaska Commissioner of Education and Early Development relied on an AI tool to draft a policy on cellphone use, and the resulting content included citations to academic studies that do not exist.

Van Buskirk called for the policy to be more descriptive as to exactly what uses of the technology are and are not allowed. The broad language proposed leaves unclear which students will be using the technology and how.

“We need to be very, very clear so students and teachers know the rules.”

Dendurent said she didn’t support putting strict parameters on any AI policy, especially as the technology is “evolving.” She said that a revision to some of the language could be considered next month.

Board President Jason Tauriainen agreed with Dendurent, but said that the broader suite of board policy already prohibits misuse and misconduct by students. If the concern is students’ copying and pasting generative AI responses rather than doing their work, plagiarism is already defined and prohibited.

Tauriainen said discussions of AI use will extend far beyond the initial policies now being considered and targeted for implementation ahead of the start of the coming school year.

“I see this policy being worked on yearly, at a minimum,” he said.

The KPBSD primarily uses two generative AI technologies designed for use in education, Holland said during Monday’s meeting: a special version of Google’s Gemini and MagicSchool. Addressing data privacy concerns,  Holland said both are specially licensed and “are not open to the internet.” The district’s version of Gemini is not connected to the public version of Gemini accessed by a standard Google search. 

Potential uses of AI for KPBSD staff, the superintendent said, could include using tools to modify or generate altered materials to meet the different needs of students.

Each of the policies, including ones that members of the committee seek to amend, will be considered again at a work session of the policy review committee in August. That meeting has not been scheduled as of this week.

A recording of the meeting and the text of the policies can be found at egov.kpbsd.org.