AI-Writing Detection Tool in Canvas
SimCheck (a Turnitin product) is an integrated plagiarism tool within Canvas. It provides both an Originality Score (percentage of submitted work that is found elsewhere) and an AI Score (percentage of submitted work that is potentially written by AI). Both reports can be accessed from within the SpeedGrader view. Please note that instructors and students can view the Originality Score, but only the instructor can view the AI Score.
Outside of Canvas, there are stand-alone AI-writing detectors you can use. Please note that these free tools are often limited in how many words they can check. If you have a lot of text to check, you'll need to do it in small batches.
GPTZero - Limited to 5,000 characters per detection
ZeroGPT - Limited to 15,000 characters per detection
Caution: At the moment, AI-writing detection tools are not entirely reliable. There are many "false positives." If a piece of writing is flagged as "AI-written," further investigation is warranted. Watch the webinar on AI and Academic Integrity for a detailed discussion (see below).
AI detection tools work by recreating questionable text using AI models. If the questionable text has high variability, it's likely that it's not AI written. However, there are "remover tools" that can "scramble" or "wash clean" AI writing, making them virtually undetectable by traditional AI detection tools. Some of the most popular AI remover tools include Undetectable AI, HideMyAI, and WriteHuman. If you come across writing that sounds machine-generated but has a 0% AI score, then most likely an AI remover tool was used. There are second-generation AI writing detectors that can detect "sanitized AI writing," but it's a matter of time before new AI removers will come onto the scene.
Here are some recommended practices for how to deal with suspected AI-related academic integrity violations:
Compare the writing in question with the student's previous work. Does it match their style, tone, and level of previous writing? If there are significant differences, it’s worth further conversation with the student.
Check for inaccuracies in sources, arguments, facts, etc. AI text generators often have no concept of something being accurate or truthful, so they may often “hallucinate” and generate fictitious sources, non-sensical arguments, and/or utter falsehoods.
Meet with the student who is suspected of misusing AI. Focus the meeting on “seeking to understand,” instead of “proving who’s right.” Ask the student how the paper was written, what was the process like, and what additional tools, if any, were used. Asking these “writing process questions” will often reveal important insight about the paper's integrity. Also, clarify with your students what types of AI tools may or may not be used, such as writing assistant tools like Grammarly and Quillbot that can summarize, paraphrase, and/or suggest writing style edits.
If this issue is prolific, discuss the issue with the entire class. If you find multiple students with high AI scores and/or suspect inappropriate AI use, then it’s worth having a conversation with the entire class, directing them back to your AI classroom use policy. If you don't have one, implement one immediately (see resource).
Offer a second chance. Consider allowing students to redo the work without penalty or for a reduced grade. If it is clear that the student used AI-generated text and did no original work, give the assignment a zero or a substantial reduction in grade.
If a violation has indeed occurred, it’s essential to establish what progressive discipline might look like. For example, a first offense results in an oral warning and an automatic failure of the assignment. A second offense results in filing a violation report and an automatic failure of the assignment. A third offense results in filing another violation report and failure from the class. Please also refer to the APU’s Academic Integrity Policies for details on handling violations.