In preparation for the Trent Central Student Association (TCSA)’s Fall By-Elections, each year’s candidates are asked to prepare biographies and campaign platforms to share on the TCSA’s website. 

Prior to the onset of this year’s electoral shitstorm, however, Arthur noticed  troubling tendencies in Electoral candidate biographies: six out of eight flagged as being AI-generated on four different AI detectors.

Arthur was alerted by the robotic tone of certain biographies, and elected to test them for AI generation against the work of its own writers, finding no traces of AI in our own staff’s writing, and from 23% to 100% AI generated text in the TCSA candidates’ profiles.

Following up on this investigation, we inquired with the Association’s Electoral candidates and Lead Electoral Officer about these findings in hopes of some clarification. 

The first to respond was ex-TCSA President and strict follower of Electoral Policy, Riya Jaykar.

In her email to Arthur, Jaykar explained that she used Microsoft Copilot, a "tool” that helps her “message to be clear and concise so that [she is] not misunderstood in any form.” As a third-language English speaker, this added clarity felt important to her, a point later echoed by Lead Electoral Officer Wendy Walker. Jaykar also claimed that Copilot had been “authorized by Trent University for the use of its students.”

Lead Electoral Officer Wendy Walker’s response to our inquiry provided further clarity, explaining that for the TCSA, “AI is a new tool and is something we are embracing rather than a flat-out ban. The Elections Committee has ruled that the use of AI is allowed for candidates but should only be utilized as a tool, not a replacement for candidates' original ideas and goals.”

She identified AI as a “beneficial source” for “International students and ESL learners, who may face a greater disadvantage in campaigning effectively in a non-native language.” 

Due to the admittedly unreliable nature of AI detectors, the TCSA elected to leave the work of evaluating candidate biographies for “originality, feasibility and practicality” up to the Lead Electoral Officer herself. These measures were, in Walker’s words, implemented “in consultation with Trent's recently updated policies regarding AI and its use in academic spaces.”

The assessment of “originality, feasibility and practicality” does not seem to follow any formally established guidelines. The primary means through which Walker makes this assessment is her own experience as Lead Electoral Officer and Association Resource Manager, which allows her to “understand what would be "feasible" for a student to accomplish, factoring in limitations such as time, money, people, and other responsibilities beyond students' leadership roles.”

Walker also claims to “review all TCSA policies so I can advise [candidates] on what would or would not break our policies.”

The policy in question indicates that “[for] any and all campaign materials, they must be original”, and that any use of “new technologies” must be approved by the Lead Election Officer. To what extent work that is at least partially generated by Artificial Intelligence can be considered original is up to the reader to decide.

Trent’s own policy on AI, which Jaykar and Walker previously alluded to, allows the use of Generative AIs such as Copilot, so long as it follows these five principles.

1. Choice: it is up to instructors to choose whether or not AI is permitted in their courses.
2. Transparency: AI use must always be disclosed by professors and by students.
3. Ethical Use: students & professors are to consider the “broader ethical implications” of their AI use as it relates to academic integrity standards, “storage and management of personal data and intellectual property”, job security, and consideration of potential environmental impacts.
4. Meaningful Learning: AI use must be aligned with Degree expectations, and generally enrich the learning experiences of students.
5. Evidence-Based: Trent is committed to seeking and incorporating feedback regarding its AI use.

While it is made very clear that AI use must be explicitly allowed by the relevant instructor, lest it constitute an academic policy violation, one is left with little guidance in situations where AI has been allowed.

This policy does demand an understanding of the “ethical implications” of AI as relating to its impacts on intellectual property, the environment, and the loss of certain work positions, but it does not ask AI users to try to mitigate these harms. It seems that a mere acknowledgement that harm is being done substitutes any efforts to reduce that harm.

AI must be enriching and “aligned with Degree expectations.” Without a clear definition of what is ‘enriching’, this point leaves much up to the individual interpretations of each administrator.

Finally, Trent pledges to listen and learn to its students and community about AI use. While describing its policy as a collective learning process is a fantastic failsafe against potential consequences of its abstract nature, it is unclear what Trent is doing to seek its students and community’s opinions on the matter.

Trent and the TCSA’s choice to allow AI in certain contexts is not the subject of debate here. Rather, what might raise concern is the fact that this AI use lacks any concrete boundaries. Aspirational guides to abstract values are aplenty, but where exactly is the line between ‘assistive’ AI use and plagiarism?\

Editors Note: This is a developing story. As of Monday October 28th, Trent University has issued a survey on the use of Generative AI to all students. Check in regularly at TrentArthur.ca and subscribe to our newsletter, The Courier, to follow this story as it develops.