Practical Responses to Generative AI

Since OpenAI first rolled out its ChatGPT to the public late in 2022, the generative AI market has grown and diversified. In addition to OpenAI’s releasing another large language model, new tools appear in the field constantly. At the same time, concerns are multiplying about the risks generative AI may pose to education, industries, and internet user safety.

In education, a number of quick solutions has been developed, such as AI-detecting functionalities and GPT teach-outs. But as we learn to work around the increasing threat of hard-to-detect plagiarism, we also face new questions: how can we incorporate generative AI into the classroom? Can it be a teaching and learning tool? how can we teach students to interact with AI safely and usefully? how will the advance of generative AI shape our students’ professional futures, and how should teachers respond to that?

Generative AI Landscape

For a comprehensive list of AI tools relevant for higher education, with key features, pros, and cons, see Generative AI Product Tracker maintained by Ithaca S+R.

Below are a few highlights identifying some of the best-known text and image AI tools. Note that no AI tools are currently under contract for Montclair users, and this list is not exhaustive.

  • ChatGPT “is a variant of the GPT (Generative Pre-training Transformer) language model, which was developed by OpenAI. GPT models are trained to generate human-like text by predicting the next word in a sequence based on the words that come before it. ChatGPT is specifically designed to be used in chatbots and conversational systems, and it is trained on a large dataset of human conversations to learn how to generate appropriate responses in a variety of contexts” (response from ChatGPT on 1/3/2023). This model is also known as GPT-3.5. Learn more:
  • GPT-4o: the newest version of OpenAI’s language model systems, officially launched in May 2024. It is a multimodal model, in that it accepts and generates audio, text, and images. It has demonstrated much stronger academic performance compared to the GPT-3.5 model. Learn more: OpenAI’s launch announcement on GPT-4o.
    • ChatGPT-4o is only consistently available through a paid ChatGPT+ (or Team) subscription at this point (starting at $20/month.)
  • Copilot (formerly Bing Chat): Microsoft’s AI chatbot. In both the free and subscription versions, as of March 2024, users have access to GPT-4 Turbo. Learn more >>
    • Microsoft’s AI-powered Designer image generator is integrated into Copilot.
  • Gemini (formerly Bard): Google’s AI chatbot. Currently not available under @montclair.edu Google accounts.
  • Adobe Firefly: image generator licensed for Montclair users (sign in with your Montclair email)
  • Dall-E 3: ChatGPT’s visual creation sister, also run by OpenAI. Available only by subscription.
Teaching with Generative AI

Generative AI’s impact on higher education extends far beyond plagiarism concerns. While plagiarism and ethics issues are very real, the advance of generative AI also created an opportunity for educators to focus on the challenges of cheating and work out strong solutions. Another important consideration is the changing job markets. For many fields and professions, digital fluency and proficient use of generative AI have become essential competencies that employees are expected to use every day on the job. For these reasons, it is crucial to integrate the basics of AI literacy in college curriculum.

Currently Montclair does not offer in a centralized way synchronous or asynchronous modules providing students with foundational knowledge they need to use AI safely and effectively, although that work is underway. However, instructors are encouraged to implement course policies and pedagogical approaches that promote AI literacy among students and assess student learning in plagiarism-resistant ways.

Tips and ideas for promoting AI literacy:

  • Develop a course policy on AI use, addressing not only plagiarism concerns and ground rules, but also issues of AI ethics, authorship, privacy, and online safety. Discuss it with students on the first days of class. Continue discussing the relevant aspects of the policy throughout the semester.
    • Keep in mind that students are likely facing major differences in faculty opinions about generative AI. Whether or not you allow generative AI in your classroom, be very clear about your expectations.
  • Teach your students to use generative AI safely. Most companies that own generative AI tools reserve the right to maintain and utilize user data for various production purposes, as well as share them with third-party stakeholders. Additionally, for the majority of tools, the company will own any outputs the tool generates. This has far-reaching implications for the definition of authorship and public knowledge domain. Additionally, there is often a possibility that small batches of user data may undergo human review in the production process within the tool-owning company. For these reasons, teach your students to never upload unpublished work or share personal and sensitive information within generative AI tools of any kind.
  • If generative AI tools are allowed in your class at all, teach your students to cite generative AI correctly.
  • Ask your students to examine AI outputs critically.
    • For example: to develop your students’ critical thinking skills, ask them to generate a ChatGPT response to a question of their own choosing, and then write an analysis of the strengths and weaknesses of the ChatGPT response. Discuss the findings as a group.
  • Encourage students to learn more about ethical uses of AI: personalizing learning, enhancing productivity, streamlining routine tasks that do not involve concerns of privacy or authorship.
  • For more ideas and details, see Teaching with ChatGPT: Assignment Design Tips and Ideas.
Mitigating Cheating and Non-Learning

Using Generative AI without proper acknowledgement to create or enhance submissions when an assignment does not explicitly call for it is academically dishonest. It is not, in fact, fundamentally different from having another person write your paper, take your test, or complete your assignment. While it may be difficult to come up with assignments that are completely AI-proof, the tips below can help you make your classroom more resistant to plagiarism.

  • Talk to your students about academic integrity, broadly. The first step to addressing the projected or actual academic integrity issues in the classroom is to talk to your students about your expectations. Remember, the policies in their other classes may be vastly different from yours!
  • Remind your students of the University’s Academic Dishonesty policy
    • Give examples. Be specific and frank about your concerns.
    • Raise questions to stimulate reflection: why is academic integrity valuable and important to uphold? What’s the point of pursuing a degree, of taking a class, if you don’t learn?
    • See Academic Dishonesty and Student Cheating for additional guidance.
  • Make plagiarizing difficult. Use some of the assignment design strategies suggested here to encourage honest work.
  • Use social annotation. For short reading responses, instead of using discussion boards or other forms of written answers to open-ended questions, try social annotation tools that require students to engage with a text along with their classmates. Try Hypothes.is or Perusall, both of which are supported by the University.
  • Ask students to use diverse media. Replace an essay or short-answer writing assignment with one that requires students to submit an audio file, podcast, video, speech, drawing, chart, diagram, or multimedia project. While artifacts can be generated with AI for every media, AI-generated images and videos are typically easier to identify than AI-generated texts.
  • Create connections to real-world experience that AI will not have. Many AI tools have a knowledge cut-off date; therefore, they will not have access to information published after that date. As LLMs rely on statistical probabilities when they generate texts, they also will underperform when asked about local events and issues. To make plagiarizing difficult, connect assignments to very recent events or new conversations in the field; to issues specific to the local community, or to discussions that took place in your own classroom. Alternatively, ask your students to find a connection between course concepts/topics and their personal experience or knowledge.
  • Run your assignment through ChatGPT or any other widely available tool. If you assign a task that can be solved by ChatGPT/ other generative AI, run it through ChatGPT first. Review the answer you receive, and tell your students about your experience (and that you’ve saved the output). ChatGPT does not produce the same answer each time the same question is posed, but the outputs may still be fairly similar.
  • Be on the lookout for AI-produced texts.
    • Currently, the University does not license or endorse any automatic AI detection tools, due to unreliable performance and bias concerns.
    • Watch out for these red flags.
    • Always consider students’ writing history and the broader context of the assignment before making a decision. When plagiarism is suspected, talking to the student individually is the easiest first step to addressing the problem.
Resources and References

For additional resources, see AI Writing and Creating Bots (Montclair netID required).

Last Modified: Monday, July 29, 2024 3:04 pm

CC

Creative Commons License
Teaching Resources by Montclair State University Office for Faculty Excellence is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

Third-party content is not covered under the Creative Commons license and may be subject to additional intellectual property notices, information, or restrictions. You are solely responsible for obtaining permission to use third-party content or determining whether your use is fair use and for responding to any claims that may arise.

Creative Commons CC BY-NC-4.0