University of Alaska Delays AI Plan Partly Written by Claude
University of Alaska regents postponed a vote on a four-page AI document until February after hundreds of faculty objected that Anthropic's Claude chatbot helped draft it. Nearly 200 students and staff petitioned for a human-written statement instead.
The University of Alaska Board of Regents has postponed a vote until its February meeting on AI Engaged, a four-page document setting principles for artificial intelligence use across university campuses that was partly drafted by Anthropic's Claude chatbot. The fight shows how far apart staff remain on what is artificial intelligence's proper role in teaching.
The University of Alaska Anchorage Faculty Senate also postponed votes on AI-related measures last week.
Why faculty object
Hundreds of faculty oppose the document because Claude helped write it. A group of nearly 200 students and staff, mostly in Anchorage, signed a petition asking regents to adopt a short statement written by humans instead.
Ben Shier, the university system's chief information technology officer, created the first draft. He prompted Claude with human ideas about appropriate AI use, after regents directed him in May to begin preparing the document.
"It was a combination of our own suggestions to the board, presentations, content, notes, as well as some notes from the board directly back to us in that meeting, all fed to Claude," Shier said in an interview Tuesday.
Shier said only about 20% of the initial two-page draft survived several rounds of human edits over the summer. He shared the result with select faculty and students days after prompting Claude.
Several people familiar with the document said Shier's use of AI was discussed internally, but it was not attributed until last month. Opposing faculty say that delayed disclosure is their main issue.
What the document does
AI Engaged does not establish policy. It requires system-wide coordination to develop, and report to regents on, a system-wide AI policy within one year of adoption. Each of the three universities must also write its own policy governing AI use.
The document also says "UA will accelerate adoption and expand capabilities across the state." Shier said it "does not bind the institutions long term to any direction."
Joel Potter, an associate philosophy professor at the University of Alaska Anchorage, co-wrote an opinion piece criticizing the document as too pro-AI. He said most faculty were shut out of editing because they were on summer break while it was written.
"It prescribes an embrace of the curriculum for the sake of student success and it doesn't acknowledge the need to resource maintaining academic integrity," Potter said.
Other university employees, however, produced years of meeting minutes that include discussions about AI policies.
Cheating and the delay
Academic integrity statistics from the university show that students' use of generative AI tools to cheat has increased since 2022, as the tools have become more common.
University system President Matthew Cooper told regents at Friday's meeting that officials had heard "many concerns," and also "real strong support for moving, and moving with urgency." He added: "It's a situation where we're never going to have everyone on the same page with AI."
Regents delayed the vote to allow more input. In the meantime, they directed the AI Coordinating Council to begin crafting campus-specific AI policies.
