Faculty Guidance for Academic Misconduct

• Dealing with misconduct should be mostly a matter of prevention.
• Have conversations early and often about academic integrity and the ethical use of technology. Build Literacy in AI (Artificial Intelligence) whether you plan to adopt it or not.
• No silver bullet exists for dealing with misconduct. Most AI detectors are hit or miss.
• The best source of information about the origin of an assignment is the student who submitted it.
• Whenever students turn away from an assignment, drive them back--deeper into the learning experience.

Pledge of Original Work 

In most LMS's, faculty can set release conditions on an assignment. Students can be required to complete a Pledge of Original Work—a one-question quiz allowing a yes (100%) or no (0%) response: “I certify that this assignment represents my own original work and that any sources have been properly acknowledged.” Such a pledge may deter some student misconduct. Answering “yes” releases the link to upload a completed assignment. 

Such a statement may also be included in the test instructions before a quiz/test in the LMS: "By submitting the quiz, I certify that I have completed it independently with no assistance besides resources specifically authorized by the instructor."  

Similarly, ProctorU requires students to sign off on the Terms of Service (as provided by the instructor on the exam form) so faculty can include a statement there as well.

Syllabus Statement 

• Be sure that your syllabus defines academic misconduct in your discipline and for your course.
• Students may have varied experiences with using AI in other courses.
• Syllabus Part B won't do this for faculty.

See Suggested Syllabus Language page.

Sample Syllabus Quiz 

• Be sure that students read your syllabus policies. 
• Early in the course, students can be required to complete a syllabus quiz in the manner of an End-User License Agreement (EULA). Making this an orientation assignment can contribute RSI requirements: I understand that I may not use artificial intelligence to generate any final products that I submit for credit as my own original work. I also understand that AI is a quickly evolving field, and if I have any questions about responsible use of technology, I should seek guidance from my professor.
• Answers for this question might be yes/no, or they could allow for a scaled response, enabling a faculty member to intervene if the response isn’t 100% certain.

Class Activities 

Faculty have the option to be “above board” with students concerning the use of AI tools. Discussion about their use and rhetorical implications are likely to lead to considerations of ethical use and higher-order thinking on Bloom’s Taxonomy. 

• Ask students to use AI to generate essays on topics / genres the class will cover. Then have students critique the results as a primer for peer review of classmates’ actual essays or as a way of understanding the goals of the assignment. Together, following analysis of AI output, students and faculty might build a rubric for assessing the class writing assignment, thereby demystifying the assignment and grading processes and promoting student success.
• Citation formats exist in both APA and MLA documentation formats for AI tools. Faculty can discuss with students the rhetorical implications of citing a chatbot as a source on a research paper or other professional work.
• Faculty can ask students to search for and rate AI-enabled apps regarding how useful or ethical they appear to be.

Additional Prevention for Courses Requiring Writing 

• For any discipline, require students to save all drafts. Tell them that even wrong turns are part of the journey and learning which way to go.
• Option: require students to use a cloud-based word processor with document history enabled.
• Require that students complete X number of process components from a list--proposal, idea web, outline, annotated bibliography, etc.
• All of this becomes a process portfolio that students submit with their final copy. (Faculty can "grade" it by bulk or completion.) This document history then becomes an asset in cases of uncertain authorship.
• Require components that AI cannot easily create (charts, images).
• Drive the student deeper into the learning experience.

Research Journals 

Have students keep a Research Journal. (Faculty could provide a worksheet.) Students will have to track:

• Databases searched (required + elective/optional)
• Search strings and types (keyword, subject, Boolean, etc.)
• Citation searches based on substantial works involving the subject.
• A log of the student’s results & decisions (keep or reject + logic)
• CRAAP Test (Currency, Relevance, Authority, Accuracy, and Purpose)
• Info about authors (preliminary to an annotated bibliography)

Unicorns and Other Clever Traps 

• If students engage in flagrant misconduct, they will tend to do minimal work. This might mean copying an assignment prompt without reading it and pasting it into AI without reading either the prompt or the output.
• Some faculty have included white-colored (“invisible”) text with unlikely words that an AI will tend to proliferate; including the word unicorn at the end of a line is one example. This works for both writing prompts and quiz/exam questions.
• If faculty find unicorns in odd places in the student’s assignment, this oddity may serve as additional proof of misconduct.

Suggestions for Online Testing 

• Faculty need options for online testing, not one-size fits all.
• Online proctoring is one option; however, it may incur additional costs, anxieties for students, and implications for the testing environment. 
• On-campus testing (at campus testing centers) should be an option for students who have documented, substantial reasons not to use online proctoring. The syllabus should direct students to the corresponding Dean who may grant an exception.
• As one possible alternative, faculty may require students to walk through their process in solving one problem on an exam. The medium could be Flipgrid, Webex or some other video-enabled platform. Such a policy will promote academic integrity and effectively measure student learning.
• Turnitin does not integrate with short answer/essay quiz responses (or discussion board postings). Therefore, faculty may consider using a Turnitin assignment (i.e. upload container) to deliver such questions, thus allowing automatic AI / plagiarism detection.
• No matter the input, Turnitin has a 300-word minimum threshold for generating an AI Writing Analysis to avoid false positives. As an alternative, faculty can combine multiple suspicious postings by the same student. For shorter responses, faculty can repeat (i.e. double-paste) the student response into a file to make the 300-word threshold. Since AI detection focuses mostly on determining whether the next word/phrase matches what is predicted based on the previous phrasing, doubling the text should not skew results significantly.

• Online AI detectors are unreliable. 
• turnitin.com might be better since the company has accumulated decades of student essays prior to ChatGPT for norming.
• Resist making any misconduct determination upon one piece of evidence.
• For instance, a faculty member may notice a paper uses research without indicating sources, has robotic word choice, tends to be general, and doesn't match a student's prior work. Uploading the file to Turnitin gives the determination: "100% AI Writing."
• Note, however, that Turnitin always includes a disclaimer: Caution: Percentage may not indicate academic misconduct. Review required.
• Interestingly, Turnitin uses language similar to that found on most vitamin supplements: This statement has not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.

Faculty can Recognize AI Traits on Their Own

Minor Traits 

• A hackneyed transition into the last paragraph: "In conclusion, ..."
• Language that seems sophisticated and cold or sounds like a corporate brochure. 
• Buzzword business-speak typifies early AI models: for instance, "Testing out novel methods and modalities."
• A tendency to generalize (nominalizations, people with titles but no names)
• Gratuitous dialogue (seems randomly placed, offers no clarity or emphasis)

More Significant Traits 

• Users can ask ChatGPT to make a Works Cited page, but the sources may be engineered, not real.
• ChatGPT cannot coordinate signal phrases and Works Cited entries.
• Factual errors (hallucinations): The AI may add details that fit the general context based on probability, not truth.
• The AI is trained on what people know: common knowledge and hot topics may be accurate while finer details may not.

Have a Conversation with the Student 

• If a faculty member suspects that student work is not entirely original, avoid confronting the student with accusations: “I know you used AI to cheat on this paper.” Unlike regular plagiarism detection, AI detection is not very reliable. 
• Instead, ask suspected students to explain their processes in writing a response or solving a problem. That dialogue will be productive either in demonstrating the student’s competence (and removing doubt) or in revealing gaps in the student’s learning. 
• If the student cannot explain the process, then the student will realize that learning objectives have not been met; if misconduct is involved, this will be a humbling and potentially beneficial experience for the student.

Challenge the Student 

Faculty who suspects academic dishonesty in writing can challenge the student’s methods or product in other ways that may be used to deepen the learning experience, create deterrents, and deal with misconduct. 

• Make a quiz using significant/suspicious words from the student paper.
• Ask about how the paper was written.
  • Tell me about your process for writing this paper. What did you do first? (Then what did you do?) 
• Ask the student to explain a claim or line of reasoning in a section of the paper.
  • Can you elaborate on "the hegemonics of shoegaze metal"? This sounds like an intriguing idea. 
• Ask the student to show how a particular research source was found. (This may be particularly helpful if faculty specify the database students must use—e.g., Academic Search Complete instead of just Googling it.)
  • I notice that you refer to "various studies" in paragraph 3. What source(s) did you use? How did you find them? Can you show me? 

Assign the Grade Earned--or Withhold the Grade

• Although AI detection may be uncertain, faculty should not hand out passing grades for questionable work. After all, a grade dispute policy exists for students who wish to pursue it.
• Faculty should not feel compelled to issue a grade. An override grade like “Need to discuss” is one option.

Be Transparent and Treat the Misconduct as a Teachable Moment 

• Address your students or post a news announcement telling the class that AI-based misconduct has occurred. Explain the consequences. Don't be an accomplice in keeping the situation a dirty little secret.
• This announcement functions like anonymized public shaming, which acts as a deterrent. (Note: Do not engage in actually calling out the individual student or making accusations in front of the class--whether in-person or online.)
• No matter what has happened, be positive and supportive. Tell the rest of your students (the majority) that you appreciate their earning honest grades. Tell students that most poor choices are not premeditated but often happen because of pressure, competing responsibilities, and limited time. Therefore, remind students of any extension policy that you may have. Invite them to ask questions in class, use office hours, or email with any uncertainties. 
• Once lost, trust is not easily regained.

Report All Cases of Misconduct  

• Use the Student Concerns card in My Tri-C Space to report student misconduct.
• Faculty should always report issues of misconduct to the Dean of Student Affairs Office since the student may commit (or may have committed) academic dishonesty in other courses. 
• When making a Student Concerns Report, be sure to include the assignment at issue, your comments, a screen capture showing AI-generated text (or an originality report indicating plagiarism), and your syllabus.
• Reporting the incident allows a report to exist and records how the incident was handled. This is particularly important in cases of repeated misconduct.

Avoid Emotional Fallout: Play the Right Game

• When one student engages in misconduct or sends suspicious assignments, faculty may become angered or obsessed with finding out the truth or trying to force a confession. 
• Those efforts will take time from other students, drain faculty of energy, potentially lead to cynicism, or damage one’s sense of goodwill towards humanity at large. 
• Dealing with misconduct can seem like a game, but don’t let it become poker by forcing the student to show a hand or admit bluffing. 
• Instead, realize that this is a role-playing game. The student may be assuming the role of author for work that someone or something else produced. Force the student into that role: make the student own the work. Demand the students to show rough work or process writing. Require revisions.
• If conditions aren’t met and faculty have more than one reason to suspect misconduct, faculty should consider assigning a consequence rather than worrying about finding the truth.

Remember that you are not alone if you suspect misconduct and want to discuss next steps. Reach out to your campus community for support: fellow faculty, your coordinator, your dean, instructional design staff, and faculty development. 

We can safely say that students may make poor decisions without being bad people. Similarly, faculty are likely to encounter student misconduct without that indicating fault in the faculty member's course design or pedagogy. Yes, best practices and emerging methods of prevention exist. However, faculty can implement all of those measures and still face misconduct; it's part of the job. In the best of cases, student misconduct can be an opportunity to learn and improve--for everyone. 

Consequently, faculty--whether younger or older, full-time or adjunct--should not feel shame or be afraid to reach out for support during cases of suspected or confirmed misconduct. As college faculty, we are necessarily engaged in collaborative work; as community college faculty, that collaboration is even more essential for our students' success. In fact, the AI Taskforce itself embodies that principle of sharing, collaboration, and working through solutions. No one should feel compelled to figure this out in isolation. No one should feel alone in trying to do what is right--especially when the best options aren't always clear at first.