Evaluating Student Work When AI is Suspected

• Is an AI policy present and clear in the syllabus and/or assignment?  
  • If not, you can still pursue an intervention or violation report based on University plagiarism policies; however, consider adding clarifying language. 
• Are you checking the work of students in a fair and equitable manner?  
• If using an AI detection tool, is it possible there was a false positive? These are more likely in certain situations: 
  • Student Use of Editor Tool Assistance (e.g., Grammarly, Microsoft Editor, etc.) 
  • Small Sample Size of Written Work (less than 300 words) 
  • Students with Accommodations 
  • Non-native English Speakers

Academic Technology & User Services (ATUS) and the Center for Instructional Innovation & Assessment (CIIA) recommend a multiple-measures approach when evaluating student work. Educators are the closest to their instructional content and their students. Evaluation, therefore, should begin with instructors using their own best judgment about the student's writing.

Carefully examine student work. Remember, while these suggestions may be indicative of AI generated content, they may not be definitive.

• Is work consistent with previous work samples? 
  • Writing style, quality, and quantity
  • Writing mechanics (such as spelling, punctuation, or structure) 
• Does the work include any of these "red flag" characteristics?
  • Not consistent with assignment guidelines. If an assignment asks students to reference specific data or sources, an AI system may not be able to follow the directions accurately. Look for data and sources in the assignment that are not expected.
  • Incorrect Information. AI generation systems work by predicting the next word based on previous content. As a result, these systems may make up statements that seem possible, but are factually wrong. Incorrect facts in AI writing are called “hallucinations.” 
  • Incorrect Citations: AI writing generators are often unable to cite their sources accurately. While new methods of obtaining accurate sources are becoming available, one red flag might be that a large number of citations are made up.
• When placing your assignment prompt into an AI text generator, does it produce similar results? 
  • TIP: Try putting your prompt into ChatGPT before you publish your assignment. Generative AI will not produce the same results each time, but you can get a sense of the output variations it produces.
  • ChatGPT - create free account to access
  • Bing Search - available on Chrome-based browsers
• When placing portions of student work in an AI Detection tool, does it yield a high result?
  • NOTE: Student-identifiable data should not be used/submitted in non-licensed and non-WWU-sanctioned tools.
  • Please be aware that AI detection software is not perfect and may yield false results.
  • These products are not officially licensed by Western; only portions of student work that has been anonymized should be tested:
    • GPTZero - free AI detection tool
    • Origin - free browser plug-in by GPTZero that allows for easy checking of sections of work in Canvas discussion board posts up to 5000 characters.

Important Notes

• Do not share personally-identifying information: Caution should be used when choosing a third-party AI detection tools to avoid disclosing any protected information. Work submitted to these tools should be fully anonymized, per Interim Generative AI Guidelines from the State.
• Work flagged by AI detection tools is not definitely AI-generated. Due to the rapidly changing nature of generative AI, there is no foolproof method of detecting AI plagiarism.  
• Always follow your research with a non-accusatory conversation with your student. Non-native English speakers and beginning writers might use simple sentence structures or phrasing and tone that differ from standard English. This can cause AI detection tools to mistake these as attempts of using AI writing tools. 

Starting in Canvas

You can enable the Simcheck plagiarism detection tool built into Canvas assignments, but this does not detect AI specifically and it is not available in other areas where students write, such as Discussions and essay answers in Quizzes. 

As long as the student work is fully anonymized, selections from the student’s submitted work (either via copy-paste or a complete file) could secondarily be put into other available AI-detection tools, depending on their capabilities. 

Third Party Options

Free versions are very limited. You will most likely need to incur a cost to continue using a tool past its most basic use or limits based on characters, words, and/or credits. 

It can be helpful to use several tools to gain a better understanding on evaluating the submitted work and identify key areas that may need further scrutiny or clarification from the student. WWU does not endorse or have licenses for these tools; however, the following products rise to the top in our research in Learning Systems.

• GPTZero
  • Origin by GPTZero (Chrome browser plugin)
• Originality.Ai
• Winston AI
• CrossPlage by Inspira

Support at WWU

• To request AI detection consultation, contact WWU's Academic Honesty office or ATUS Learning Systems staff via CanvasHelp@wwu.edu. 

• Talk to All: You might begin by talking to the entire class about suspected AI use by a subset of the class. Explain why AI is not appropriate for this class or assignment, and what the expectations are for student-written assignments.  
• Meet with a Student: If you reach out to an individual, let them know you have a reason to believe their work includes AI, but that you would like to hear their perspective. If needed, have them explain their process on the assignment.  
• Consider It a Learning Opportunity: If you find that the conversation has addressed the concerns, you might find a resolution that does not impact the student’s grade. This might be a good time to add, amend, or improve a class AI policy. 
  • If the intervention does impact the student’s grade, be sure to do a violation report with the Academic Honesty office. This protects all parties, provides a neutral third party for evaluation, and potentially gives students access to training materials. While a report does not go on a student’s official records, repeat violations may have more serious consequences. 

There may be cases where the evidence is unclear or when a student was using an AI assistive tool unknowingly. You may want to work with the student to determine a solution that will not have a negative impact on their grade. If the resolution involves a learning opportunity with no grade penalty, you may consider some options for the student. 

• Redo and resubmit the work without the assistance of AI.
• Redo the work in your presence during office hours.
• Offer another option or alternative prompt for the work.

Once you evaluate a student's work and determine an academic honesty violation is likely, whether AI or other forms of plagiarism, and you have given the student the opportunity to meet and discuss the situation, you will communicate to the student your decision, and complete an Academic Honesty Violation form. Even at this point, the first violation reported for the student is considered a learning experience.

It is important to follow the steps outlined by the Academic Honesty office:

Academic Honesty Instructor Reporting Steps

Remember, if you take any punitive action you must report it. Reporting the incident does several things that support both you and the student. First, this allows the Academic Honesty office to determine when events are second offenses, for which the student must go before the Academic Honesty Board. Second, when you report the incident it sets into motion a formal process that also provides the student the opportunity for representation and appeal and an opportunity to participate in training to learn from the incident.