Skip to main content
Left arrow Barnard.edu
  • Who We Are
    • Staff
    • What We Do
    • Faculty Advisory Committee
    • Student Advisory Committee
  • Events & Projects
    • Events Calendar
    • Current & Past Projects
    • Community Resources
  • Teaching Resources
    • Teaching Resources Library
    • Course Design
    • Engagement and Community Building
    • Equity and Accessibility
    • Campus Resources
  • Learning Resources
    • Learning Resources Library
    • Study Strategies
    • Special Topics
  • Services & Opportunities
    • Services for Instructors
    • Staff Office Hours
    • Submit Work
    • Pedagogy Grants
Left arrow Barnard.edu
Search All of Barnard.edu

On This Page

  • Introduction
  • Recommendations for Navigating Generative AI
  • Additional Resources & References

Generative AI & the College Classroom

Educators are acting fast to consider the implications of ChatGPT and several other generative artificial intelligence (AI) tools (i.e., technology that creates new data rather than simply analyzing or categorizing existing data). The following recommendations represent the CEP’s and IMATS' initial research into generative artificial intelligence and its implications in the higher education classroom, with recommendations for classroom activities, assignment design, and academic honesty and considerations for AI detection software and student privacy.

It is important to note that this software is very new, and our collective understanding of what it does and how it works is likely to evolve. As such, the CEP will continue to consult with IMATS and the Computational Science Center (CSC) and conduct additional research to update these recommendations. Since launching, ChatGPT is also frequently unavailable due to high levels of use. If you plan to use the tool in your classroom, you will want to anticipate the likelihood that it is not available and plan a back-up activity.

Please contact us at pedagogy@barnard.edu if you have any feedback or questions about this resource. We are also happy to set up one-on-one consultations with instructors to discuss generative AI in their course and disciplinary contexts.

Recommendations for Navigating Generative AI

If you’re considering using generative AI or ChatGPT in the classroom, or want to discuss these technologies with your students, it is helpful to know a little bit about what these tools are and how they work. Generative AI is a type of artificial intelligence that creates new content, based on what it has learned from existing content. The process of learning from existing content is called “training,” and results in the creation of a statistical model. When given a prompt, generative AI uses this statistical model to predict what an expected response might be - and thus generate new content.   

ChatGPT is a generative AI tool developed by OpenAI and released for public beta testing in November2022. Using a pre-trained language model, ChatGPT takes text prompts from users and produces responses that are supposed to mimic human writing in dialogue format. Notable features include its ability to answer questions, hold a conversation, summarize information, write computer code, and more. 

ChatGPT’s language model is trained on very large datasets of text data available on the internet, including websites, books, and other texts; these datasets are estimated to contain nearly one trillion words in multiple languages. This includes labeled data, which includes annotations set by humans that guide the language model in categorizing data correctly, as well as unlabeled data (sometimes called “raw data”), which does not contain any annotations. OpenAI improved the model's accuracy by having people evaluate and correct its analysis of the unlabeled data, enhancing its ability to produce written language.

ChatGPT has a variety of limitations that are integral to understanding and working with the tool:

  • ChatGPT has a limited scope of reference, as the data used to train the language model only includes information from 2021 and earlier.
  • ChatGPT may respond with factually incorrect information contained in its training data.
  • ChatGPT may reproduce biases that were present in the training data.
  • ChatGPT’s model may not grasp the context of the prompt, and is known to struggle with “common sense” knowledge, idioms, and sarcasm.

The following webinars on generative AI in the college context may be helpful for learning about the technology and its implications for academic integrity, assignment design, student engagement, and the future of AI:

  • "AI and Academia: The End of the Essay?" recording from the webinar hosted by Maple League Universities on January 31 at 11am ET
  • "AI FUTURES: An Interdisciplinary Conversation on Large Language Models and the Future of Human Writing" hosted by Rutgers' Critical AI on February 16 at 5PM ET

 

  • Have an open conversation with your students discussing what ChatGPT is and give some examples of how generative AI is being used in education generally. Consider what instructors and students need and want to know about how ChatGPT and related models are built, how they work, and their potential perils and possibilities (educational, ethical, environmental, labor, social, political).  
  • Depending on the learning objectives for your course, you might consider discussing a report such as ChatGPT: Educational friend or foe? about the possibilities of and debates around ChatGPT and recent pieces (e.g., ChatGPT Advice Academics Can Use Now, How About We Put Learning at the Center?, Don’t Ban ChatGPT in Schools. Teach With It, and How Do Students Feel About OpenAI’s ChatGPT) could be helpful in sparking a class discussion about ChatGPT in the higher education context. Such a conversation could then lead into a more focused discussion about academic production and academic honesty in your course.
  • Discuss students’ potentially diverse motivations for using ChatGPT or other generative AI software. Do they arise from stress about the writing and research process? Time management on big projects? Competition with other students? Experimentation and curiosity about using AI? Grade and/or other pressures and/or burnout? Invite your students to have an honest discussion about these and related questions. As stated in a recent article about ChatGPT-generated abstracts fooling scientists, solutions to the use of ChatGPT will likely be found by critically examining the pressures motivating students and scholars to use it rather than by focusing solely on preventing and detecting AI-generated work.
  • Cultivate an environment in your course in which students will feel comfortable approaching you if they need more direct support from you, their peers, or a campus resource to successfully complete an assignment. To take steps toward cultivating this environment, you might include a question about generative AI on your introductory course survey to understand your students’ potential interest and concerns about the tool. You could also include a discussion of generative AI in your community agreements, including the impact of AI-generated work on the classroom community. Additionally, take the opportunity throughout the semester to invite students to meet with you during office/student hours to discuss concerns about deadlines, especially during midterms and finals.

  • Spend some time discussing the definition (or definitions) of academic honesty and discuss your own expectations for academic honesty with your students. Be open, specific, and direct about what those expectations are.
  • Students generally express interest in joining the conversation about the 'why' and 'how' of learning. Understanding this purpose and process is beneficial to all students and can be motivating and demonstrate respect for their agency in the learning process. Consider connecting your expectations for academic honesty to the purpose of your course learning goals or assignments. Why is academic honesty important for this kind of skill building or knowledge acquisition and how will your class prepare students for developing these skills through the difficult work of learning? You might frame this conversation in terms of the skills you hope students will develop and the importance or value of these skills for future learning in the field or discipline or for their lifelong learning in general.
  • Review the College Honor Code with the students and explain any of your concerns about ChatGPT and other related GPT models in that context.  
  • Consider adding a statement to your syllabus such as, “Use of an AI text generator when an assignment does not explicitly ask or allow for it is plagiarism.”
  • If you use peer review or evaluation in your course, consider a specific conversation about the implications of ChatGPT within this process. In addition to the fact that generative AI work can constitute academic dishonesty, what are the ethical implications of asking peers to review AI-generated work? If you do not use peer review or evaluation in your course, consider ways that you can engage students in reviewing each other’s work and inviting them to actively contribute to their community of learning and the discipline or profession at large. 

 

  • Prioritize reflection and growth in the learning process. This might take the form of supplemental process reflections, a writer’s memo to accompany a final paper or project, or a response to a metacognitive prompt after any assessment or activity. For exams or quizzes, consider offering partial credit if students show their work or supplemental reflection questions where students can explain how and where they got stuck on a question or problem and why. When possible, give students opportunities to revise and learn from errors. In general, focusing on process over product will deepen students’ intrinsic motivation and can deter them from academic dishonesty. 
  • Consider reviewing your course assignments and your criteria for evaluating them to ensure that progress and process are centered.
  • The CEP’s Active Learning Guide, Creating an Engaging and Inclusive Classroom and Flipped Classrooms may offer helpful perspectives. Several active and flipped learning strategies, including in-class writing, collaborative coding, or purposeful group work (e.g., having students working in small groups in class to apply what they've learned in readings, viewings, or problem sets at home) can help to prioritize students’ creative and collaborative learning. In-class writing, brainstorming, or problem solving (e.g., debugging) can also encourage students to complete work they have already started in class and perhaps minimize the likelihood of their using or relying significantly on generative AI.
  • Share our resource on Tackling Large Assignments with students to provide guidance around how to manage time and successfully plan for final course assignments.

  • While the CEP does not recommend re-designing a course or assignment solely to prevent generative AI submissions, some (though not all) of the suggestions and assignment changes currently being proposed align with existing evidence-based practices for improving both learning and engagement and promoting higher-order thinking. If appropriate to your course, consider reviewing your existing assignments to promote growth, reflection, and academic honesty.
  • Though hand-written assignments may seem like a tempting alternative to typed papers or exam submissions, keep accessibility in mind; hand-written work, especially on timed exams, can be difficult for students with disabilities and students who use screen readers.
  • Though ChatGPT can be used to quickly summarize information, this tool is currently still prone to inaccuracies. Generative AI tools are also currently less well equipped for performing in-depth comparative analysis and synthesis assignments and assignments that draw on verifiable sources and quotations. Consider scaffolding an assignment by incorporating summary in low-stakes assignments or in-class activities (e.g., classroom discussions, compiled questions about a reading on Padlet or CourseWorks, concept maps), and then encouraging students to engage in higher-order thinking in written assignments, exams, or projects (e.g., synthesis of multiple texts, application of understanding to their personal lives or experiences, presentations, papers, or projects that build on prior research).
  • If given enough personal information about a student, ChatGPT can simulate reflective writing or writing that appears to draw upon personal experience. Asking students to relate details from a reading, lecture, experiment, or class discussion to their personal lives can engage students in the material and help them see its relevance, but bear in mind that such writing can still be simulated.
  • ChatGPT currently relies on available text on the internet up until 2021. Instructors may consider this fact when adapting their course to incorporate more recent (e.g., James Webb space telescope, Don’t Worry Darling) or less canonical texts within the syllabus or design assignments in different mediums or modes (e.g., concept map, collage, oral presentation, video, photo essay).
  • When possible, scaffold your assignments to promote revision and growth over time, with opportunities for feedback from peers, TAs, and/or the instructor. Build assignment pre-writing or brainstorming into class time and invite students to share and discuss these ideas in small groups or with the class as a whole.
  • If you and your students are comfortable engaging with ChatGPT (see privacy considerations below), consider experimenting with the tool in a low-stakes assignment (e.g., asking them to craft an essay, quiz, or code using the tool and then reflect on why and how the essay, quiz, or code could be changed to be more fully developed, factually accurate, or to run more efficiently or elegantly). These activities can also promote metacognition. Not sure if your students are comfortable using generative AI technology? Ask them about it on your intro course survey.

  • Several tools exist for detecting AI-generated content, including most recently a tool by OpenAI, the creators of ChatGPT. Others include GPTZero, openai-detector and Gltr.io.
  • It’s important to remember that these tools, too, are imperfect and the results will need further interpretation by a human.

  • ChatGPT may share account holders’ personal information with third parties as stated in their privacy policy and numerous articles have exposed how it replicates racist biases (see, “ChatGPT proves that AI still has a racism problem” and “The Internet’s New Favorite AI Proposes Torturing Iranians and Surveilling Mosques”, for example).
  • It is important to be aware that ChatGPT’s potential sharing of personal information with third parties may raise serious privacy concerns for your students and perhaps in particular for students from marginalized backgrounds.

Additional Resources & References

Brookfield, S. (2012). Teaching for Critical Thinking: Tools and Techniques to Help Students Question Their Assumptions. Jossey-Bass.

Columbia University Center for Teaching & Learning. (n.d.). Considerations for AI Tools in the Classroom. https://ctl.columbia.edu/resources-and-technology/resources/ai-tools/

Columbia University Center for Teaching & Learning. (n.d.). Metacognition. https://ctl.columbia.edu/resources-and-technology/resources/metacognition/

Fisher, A. (2011). Critical Thinking: An Introduction. Cambridge University Press.

Hogan, K., & Pressley, M. (Eds.). (1997). Scaffolding Student Learning: Instructional Approaches and Issues. Brookline Books.

Hughes, A. (2023, February 2). CHATGPT: Everything you need to know about OpenAI's GPT-3 tool. BBC Science Focus Magazine. https://www.sciencefocus.com/future-technology/gpt-3/

Mills, Anna and Lauren M.E. Goodlad (2023). Adapting College Writing for the Age of Large Language Models such as ChatGPT: Some Next Steps for Educators. Critical AI. https://criticalai.org/2023/01/17/critical-ai-adapting-college-writing-for-the-age-of-large-language-models-such-as-chatgpt-some-next-steps-for-educators/

Montclair State University Office for Faculty Excellence. (2023, Jan 17). Practical Responses to ChatGPT. https://www.montclair.edu/faculty-excellence/practical-responses-to-chat-gpt/

O'brien, M. (2023, January 6). Explainer: What is chatgpt and why are schools blocking it? AP NEWS. https://apnews.com/article/what-is-chat-gpt-ac4967a4fb41fda31c4d27f015e32660 

OpenAI API. (n.d.). https://platform.openai.com/docs/chatgpt-education

OpenAI. (2023, February 2). CHATGPT: Optimizing language models for dialogue. OpenAI. Retrieved February 4, 2023, from https://openai.com/blog/chatgpt/

Rouse, M. (n.d.). What is chatgpt? - definition from Techopedia. Techopedia.com. https://www.techopedia.com/definition/34933/chatgpt 

Stanford Graduate School of Education. (2022, Dec 20). Stanford faculty weigh in on ChatGPT's shake-up in education. https://ed.stanford.edu/news/stanford-faculty-weigh-new-ai-chatbot-s-shake-learning-and-teaching

Sydorenko, I. (2020, September 1). Unlabeled data in machine learning. High quality data annotation for Machine Learning. https://labelyourdata.com/articles/unlabeled-data-in-machine-learning 

UC Berkeley Center for Teaching & Learning. (n.d.).Valuing Process as Equal To, or Greater Than, Product. https://teaching.berkeley.edu/valuing-process-equal-or-greater-product

University of Connecticut Center for Excellence in Teaching and Learning. (n.d.). Critical Thinking and other Higher-Order Thinking Skills. https://cetl.uconn.edu/resources/design-your-course/teaching-and-learning-techniques/critical-thinking-and-other-higher-order-thinking-skills/

University of Calgary Taylor Institute for Teaching and Learning (2023) Teaching and Learning with Artificial Intelligence Apps 

What is Generative AI? McKinsey & Company. (n.d.). https://www.mckinsey.com/featured-insights/mckinsey-explainers/what-is-generative-ai 

Site Footer

at The Milstein Center for Teaching and Learning

pedagogy@barnard.edu

Departments

  • Library
  • IMATS
  • Archives
  • Zines

Teaching & Learning Centers

  • Center for Engaged Pedagogy
  • Design Center
  • Digital Humanities Center
  • Empirical Reasoning Center
  • Movement Lab
  • Sloate Media Center
  • Vagelos Computational Science Center

Copyright © 2023 Barnard College | Columbia University