Academic Integrity in Canvas

Academic integrity can be enforced in a variety of ways in the Canvas environment.

Quiz-based Assessments

• In quizzes, require an agreement to an honor code statement as the first question of the quiz.
  • How? Create a short answer question that displays the honor code statement and instructs students to type their initials in the answer space. Being presented with an honor code statement and agreeing to it establishes a baseline level of behavior for the student.
  • The "question" can be worth 0 points if desired.
• In quizzes, set a time limit for each attempt.
• In quizzes, select questions from a question bank/pool so each student sees a different set of questions.  Canvas quizzes can use multiple question banks (though each question "group" will occur separately in the quiz).
• In quizzes, shuffle the answer choices on multiple choice questions (be sure to avoid answers that say "All of the above").
• Require the Respondus Lockdown Browser for completion of the quiz.

Writing-based Assessments

Especially in the age of AI, maintaining the integrity of writing assessment can be tricky.  AI tools for completing assessments are constantly marketed to students and the allowance of AI in assignments varies from instructor to instructor and from assignment to assignment.  In all cases, clear communication from instructors about what even constitutes "AI" for the purposes of academic integrity is essential.  Those instructors who simply assert "No AI" without understanding the current AI landscape that students exist in, or without elaborating their definition of AI are setting themselves up for conflict and misunderstanding.

About the Turnitin AI Writing Report

As one of the best AI detectors available, in part because it's been trained on thousands of actual student writing examples, Turnitin's AI detector reliably detects AI writing.  However, its report cannot tell instructors anything about what tool was used to generate the AI writing. It also cannot tell what the student's intent was in using such a tool.  Importantly, the AI report often underreports AI writing in order to "do no harm" and reduce false positives.  That means Turnitin is highly confident that any reported AI writing is correctly flagged.

Learn more about the Turnitin AI report

Levels of Allowable AI Use

Writing assignments can be protected in different ways depending on the instructor's goals.

Level I: No AI

In this level, AI of any kind is not permitted for the generation of content (prose) that the student passes off as their own work. This is the hardest level to enforce because the instructor must also control the student writing environment as much as possible. It's also likely that the instructor will need to be able to review the student's writing process.

Without controlling the writing environment, students will have access to tools that students may believe are not "AI" but which nevertheless are flagged by AI detectors because they generate prose.  Grammarly's revision tools are by far the biggest source of confusion in this regard: they generate AI prose but students aren't aware that such prose will be flagged.  This leaves them potentially having to defend themselves against academic integrity charges with no evidence to support their defense.

Assignment options:

1. In-person, on-paper
   In this situation, students hand-write their content in an in-person setting arranged by the instructor.  Electronic access to supplementary resources may or may not be allowed as desired.
2. Turnitin Clarity
   Whether in-person or remotely, in this approach students compose and submit their content using the Turnitin Clarity writing environment. While Clarity does not prevent students from pasting text generated by an outside AI, such actions are flagged by Clarity and may be submitted to the AI detector (if more than 300 words). After submission is complete, instructors can access a report and playback of the student's writing process. With Clarity, students can pause and resume their work as often as necessary until the due date of the assignment.
3. Canvas Essay Question w/Lockdown Browser
   In this situation the instructor creates a 1-question quiz in Canvas using the Essay question type.  The quiz will have no time limit but students will be limited to one attempt.  Then, the instructor enables the Lockdown Browser requirement for the quiz.
   
   To write their assignment, students launch Lockdown Browser and begin the "quiz".  Without a time limit, students can pause and resume their writing as often as necessary until the due date of the assignment. Lockdown Browser requires students to provide a reason for leaving the "quiz" each time they do so without submitting, but any reason will suffice, such as "work-in-progress". Canvas saves their progress constantly.
   
   The only way to get information about the student writing process is enable the Respondus Monitor add-on and select "Screen Only" for the recording type.  This creates a simple recording of the student's screen during that particular writing session.  Multiple sessions will produce separate recordings for the instructor to analyze.
   
   Crucial caveats: As a lockdown environment, Lockdown Browser does the following:
   1. Prevents the students from viewing unauthorized websites while working.
   2. Prevents the students from opening any supplementary documents (even legitimate ones) they may need for reference.
   3. Prevents the students from copy-pasting content from outside the current text area (such as from a previously written file on their computer).
   4. Prevents the students from including images and web links in their document (other basic word processing tools are allowed).
      
      Because of its general incompatibility with modern electronic writing processes, t​​​​his option is probably best used with in-person writing assignments where students are not expected to refer to other materials while writing.

Level II: AI Feedback and Coaching

In this level of allowable AI usage, students are permitted to use AI to receive coaching and feedback on their works-in-progress. Coaching can also take the form of initial brainstorming and other pre-writing tasks with the student.  In order for coaching to be effective, instructors must include thorough instructions in the assignment prompt as well as a rubric of sufficient thoroughness as to be useful by the AI coach. This model most closely replicates an on-campus writing center/tutoring service.

In terms of academic integrity, both of the following options support AI writing detection using Turnitin.

1. Discussions Plus/Harmonize
   Harmonize can use AI to help instructors generate both the instructions and the rubric for any Harmonize activity, such as discussions and peer review.  As they work, students can send their submission to the AI coach for rubric-based feedback.  The coach is not a chatbot (no brainstorming) and cannot generate any prose (whether as corrections or additions to the student's work).
   
   Importantly, if enabled by the instructor, Harmonize can send student submissions to Turnitin's AI detector.  Therefore, instructors can access Turnitin's AI report for any content that meets Turnitin's requirements for AI detection (min. 300 words).
2. Turnitin Clarity
   In this approach, students compose and submit their content using the Turnitin Clarity writing environment. Instructors provide detailed instructions and a thorough rubric.  Instructors then enable the AI Coaching tool.  In this tool, students can chat with an AI to brainstorm and refine their work.  The AI is trained to refuse to generate prose and any text pasted from outside the Clarity writing space will be checked for AI.  Additionally, students can be permitted by the instructor to make use of Clarity's built-in spellcheck and grammar checking tools.  After submission, the student's AI chat history is made part of the overall writing report that is available to the instructor.