Dear Colleagues,
This past spring, I distributed Guidelines for Using Generative Artificial Intelligence at GW. I now wish to provide you some additional guidance as we begin the new academic year.
I encourage you all to remain sensitive to generative artificial intelligence (GAI) and its intersection with academic integrity at GW — topics addressed in my prior note to the faculty. In this message, I wish to supplement that guidance with some just-in-time insights as you prepare yourself and your students for the start of the semester as well as highlight some GAI-related resources available through the LAI Instructional Core.
Help students understand the meaning of academic integrity as well as your expectations for written products
- Reinforce what plagiarism/cheating is – why it’s wrong and your expectations for turning in original work.
- Reference pertinent policies and ethics related to student conduct.
- Consider including a declaration of honesty in your assignments where students promise to abide by the integrity standards you, the instructor, GW, and students expect of themselves.
- Allow students to disclose, if the instructor permits the use of GAI, the use of GAI tools like ChatGPT and GPT-3 and explain how it was useful in completing an assignment.
- Require at all levels meticulous citing of peer-reviewed and published work and identify the reason for this.
- Consider allowing draft submissions prior to final submissions so that faculty and instructional staff can get to know students’ writing voice and make targeted recommendations for improvement, thus minimizing last-minute attempts at using unauthorized tools.
- Plagiarism detection tools are available and should be used if you suspect a breach in academic integrity. SafeAssign is available to all GW instructors and can be found in the Blackboard suite of tools. Keep in mind that there are no tools that can reliably detect GAI-generated text and that false positives may put you in a very difficult position with students.
- Be clear in your expectations and refer to learning objectives and grading rubrics or criteria when assessing a student’s work.
- When possible, ask students to present work they have done orally or in a public forum like a class session.
Understand the limitations of GAI chatbots
In understanding the behaviors and limitations of GAI tools outlined below, you can look more critically at student responses and better detect inappropriate use of GAI to generate work.
- Chatbots can sometimes use verbiage repetitively or in unusual contexts. Look for phrases that were never used in the classroom but somehow find themselves in written work by a student.
- Chatbots cannot create new ideas but only pull from the vast data sets that exist on the internet provided by others in digital formats. Encourage creative and critical writing prompts in your assignments, asking students to perform more analysis and synthesis rather than repeat facts.
- Chatbots do not cite their own writing and can generate false citations. Require citations by students to ensure that peer-reviewed and published material from your course or elsewhere are supporting their thoughts. Look for originality in writing. Spot-check cited materials to ensure that they exist.
- Chatbots do not produce error-free content. Seek out inconsistencies and materials that do not adhere to course materials and adjacent assignments.
- Chatbots do not produce contextually aware prose (especially for the casual user). Require students to write within context using real or even fictitious scenarios to ground their thinking.
Final Thoughts and Additional Resources
Remember, there is no failsafe method to prevent plagiarism. However, you can use available tools to attempt to limit the use of GAI to take the place of original work. Your human skills as instructors continue to be your best defense against dishonesty, and education about why integrity is important to the communication of knowledge is key.
Assistance from the LAI Instructional Core professional staff is always available by contacting instructioncore
gwu [dot] edu (instructioncore[at]gwu[dot]edu).
For more information about these suggestions and the emerging chatbot landscape refer to:
- Debby R. E. Cotton, Peter A. Cotton & J. Reuben Shipway (2023): Chatting and cheating: Ensuring academic integrity in the era of ChatGPT, Innovations in Education and Teaching International, DOI: 10.1080/14703297.2023.2190148
- Ryan Watkins (2022): Update Your Course Syllabus for chatGPT
This guidance is also available on the Office of the Provost website. I wish you the best of luck as you start the semester!
Christopher Alan Bracey
Provost and Executive Vice President for Academic Affairs
Professor of Law