As new tech such as ChatGPT emerges that can create and refine content (written text, code, problems), it is important to both familiarize yourself with the tools and consider potential teaching strategies using these tools. Here, we’re offering some ways to approach the use of AI as well as an ongoing collection of articles and resources. Join our faculty forum on Feb. 9th to discuss this tool with your colleagues.

What is Chat GPT?

Most likely know ChatGPT from the headlines it is receiving. ChatGPT can be used to write clear, human-like text, produce computer code and scripts, answer some math and statistics problem sets, and more. To learn more about what it is and why some people are concerned, you may want to watch this short interview with Prof. Emily Bender and read this New York Times article. If you like, you can also read what the Washington Post editorial board thinks about potential societal impacts. Consider this list of ways to harness AI across a variety of tasks. ChatGPT is the most recent AI tool to receive attention, but other similar tools will follow (e.g., Quillbot AI).

Perhaps the most unsettling aspect of ChatGPT is its ability to provide clear, concise text based on a prompt provided by the user. Here's how ChatGPT describes itself (Jan. 3, 2023):

GPT (Generative Pre-training Transformer) is a type of artificial intelligence model developed by OpenAI. It is a large language model that is trained on a massive dataset of text and is able to generate human-like text by predicting the next word in a sequence. There are several versions of GPT, including GPT, GPT-2, and GPT-3. The main difference between these versions is the size of the model and the amount of data that it was trained on. I'm not sure what you mean by "Chat GPT." If you are asking about using GPT for chat applications, it is possible to use GPT to generate responses to user input in a chat application, although it may not always produce results that are appropriate or coherent.

Here’s more on how Chat GPT works from Open AI and an article on previous iterations of the technology.

Artificial Intelligence and Education

The release of the currently free software, ChatGPT, in November of 2022 prompted a flurry of excitement and legitimate concern. The integration of Artificial Intelligence (AI), natural language processors (recall Clippy from Microsoft), and other machine learning into our education and work processes is certainly here to stay.

While ChatGPT seems especially innovative, we have been through this cycle of technology adoption before, from calculators to Wolfram Alpha to Google translate. As with each of these tools, the technologies will only improve. Because of this, it may be best not to respond to ChatGPT alone, but rather to acknowledge that such technology will change how we write, even if we don't yet know the exact nature of that change.

Because some educators worry about a changing social contract in which students submit work they haven’t produced themselves, the first instinct might be to ban ChatGPT use altogether, which some schools have done. Instead of forbidding its use, however, we might investigate which questions AI poses for us as teachers and for our students as learners. This approach aligns with such Georgetown values as “contemplation in action,” which holds that reflection is a critical aspect of intellectual inquiry.

One way to acknowledge this is to “teach the problem.” While this may seem to be a problem only for writing instructors, all of us, regardless of discipline, should take this moment to ask “How can we think anew about producing and sharing ideas? How might the task of communication change and grow?”

Talking with Your Students About AI

We encourage you to talk with your students at the beginning of the semester about the fundamental questions AI is forcing us to ask. How do they see the role of AI om their education evolving? What are the ethical uses of AI in education and beyond?

You may consider adding a statement to your syllabus or even just consider your syllabus in light of AI tools. Different faculty will have different expectations about whether and how students can use AI tools, so being transparent about your expectations is essential. If you want to forbid using AI tools, be explicit about this on your syllabus, as with this one here. If you allow these tools but want them to be acknowledged (cited or referenced), explain that on your syllabus. See some approaches below or the policies linked here and consider including in your syllabus:

  • If you have questions about what is permitted, please reach out to me.
  • It is important to remember that ChatGPT and other AI tools are not a replacement for your own critical thinking and original ideas. The ultimate goal of this course and any tool used to submit work is to enhance your own learning and understanding, not to undermine it.
  • As a college student, it is your responsibility to maintain the highest standards of academic integrity. This includes a) ensuring that all work submitted for grades is your own original work, and b) properly citing any sources that you use.
  • Having AI write your paper constitutes plagiarism. If the source of the work is unclear, I will require you to meet with me to explain the ideas and your process.

If you’re comfortable with students using these AI tools to generate material, encourage them to focus on developing the knowledge and skills necessary to use the tool effectively. Once a draft is produced, they will need disciplinary knowledge in order to effectively ask for “more of this or less of that.” They still need expertise. AI-generation tools aren’t able to refer to your class discussion, create an infographic, handle citations, or accurately cross-reference course materials. They also can’t create a multi-modal essay—or an essay that requires sound, images, and related links, a format discussed at more length in this Tonya Howe article and a Ryan Cordell article here.

You may wish to point out that at this early and temporarily free stage of the software, AI can produce a passable but not great paper. Overall, the information is generalized, unreliable, and limited by what it has been fed. Kim Lubreski (Sociology, JUPS), looking to integrate AI into her class on refugees in DC, stumped Chat GPT by asking about the number of Iranian refugees in the area as of 2022. While the current date information feeding Chat GPT appears to extend only through 2021, this date will continue to change as the technology evolves.

Designing Assignments Effectively

As always, college-level writing should be specific, and students should be able to make their process visible. If a machine has done all the work of a piece of writing, the student has neither been specific nor could they explain their work. See below for some widely shared techniques to help achieve these goals. These suggestions move from design-oriented approaches to process-oriented ones:

  1. Design your writing prompts to make reference to material specific to your class, such as class discussions, Canvas discussions, or other unusual or even unique material. For example, use at least two theorists discussed in class to support your answer. One step further would be to ask students to refer to class discussion or their notes.
  2. Ask students to draw on their responses to each other in the classroom or in classroom contexts, like discussion boards or blogs. For example, in David Lipscomb’s “Writing and Culture” class, each student maintains a blog (via CNDLS commons), posting and responding to peers’ posts every week. Alternatively, for longer assignments, many Georgetown faculty incorporate peer review of drafts.
  3. Assign more personalized writing, the more personalized, the better. You may want to go for a mix of short personal pieces; for longer pieces, focus on editing for particular audiences and tone. This approach may also help you get to know students’ writing in more detail so that you can recognize their style and tone.
  4. Ask students to incorporate outside sources and cite them appropriately. Ensuring that these sources are integrated and referred to throughout the piece is something a student can do much better than AI (so far).
  5. Tailor your assignments to focus on a specific rhetorical context, as AI struggles to focus on the audience as well as the tone. Here is an introductory assignment in a computer science course, for instance.
  6. Have your students use ChatGPT to answer a prompt and then comment on the answer provided. Where does it succeed? Where does it fail? Where does it not understand the nuance or depth of the question (see note below about Privacy and Data Collection).
  7. Feed your assignment prompts into ChatGPT and see what kind of results you get. If GPT quickly spits out what you would consider a model response, revise your prompt to make it more specific to the class, the situation, and your particular students (see note below about Privacy and Data Collection*).
  8. Continue to reflect: What do we want students to get out of any given assignment? Where is the crux of critical thinking?
  9. Provide incentives for the process as well as the product—the behaviors and habits that are associated with strong learning. If a perfect product (test, paper) is the only way to receive an A, students are more likely to consider cheating. In your efforts to fully assess student learning, make sure to include all the good processes that are needed to be a strong learner in your course: reading, viewing, speaking, improving, responding to feedback on drafts, reflecting on one’s learning, etc. Review your grading criteria and rubrics to make sure you’re setting your students up to adopt strong learning strategies.
  10. Incorporate AI into a drafting process: 1) Have AI write a first draft and then ask students to edit it or vice versa 2) have students write the drafts and ask AI software to edit it. A key question and skill may become coaching students to coach AI to generate quality writing (see note below about Privacy and Data Collection*).

Finally, you may wish to incorporate more real-time (in-person or virtual) assignments and assessments in your course if you wish to avoid some of the potential challenges of ChatGPT in assignments.

Other resources on harnessing AI:

Interdisciplinary Considerations

One dominant line of thinking around using AI in a higher ed context is to use its affordances to encourage students to ask questions or design prompts that yield accurate, well-constructed answers. By helping students think critically about good questions, they can use the AI-produced content to build upon and focus on higher thinking. Here’s Open AI’s Prompt Design Guide, as well as a site that can help you engineer more effective prompts.

AI-generated answers may be embedded throughout the disciplines. As you focus on course content, bear in mind that students’ toolboxes for responding to course content and demonstrating their understanding have been enlarged with this new technology. Here are some discipline-specific resources:

How to Detect AI-produced Content

There are emerging sites that are improving at detecting AI-produced content. Some prominent ones include

This software is useful if the entire selection was produced by AI, but if the user has embedded some AI-crafted text within their own words, it’s much harder for the detector to identify AI. Please see the privacy notice below. For guidance in using this detection software, please visit:

Of course, as AI evolves, the detectors will evolve, as will the technology to bypass AI detectors. Some examples of these tools are CopyGenius and Quillbot AI.

Privacy and Data Collection

The use of chatbots and AI language models may involve the collection and use of personal data, raising concerns about privacy and data security. Before asking students to use ChatGPT (if you choose to do so), read over the very porous privacy policy with them and discourage them from sharing personal information on the platform. Some key points to note is that:

  • The company may access any information fed into or created by its technology
  • They use log-in data, tracking, and other analytics
  • Their technology does not respond to “Do Not Track”

Allow students to opt out if they don’t feel comfortable having their data collected. You and your students may also register with an untraceable email account if you are worried about the privacy risks. This may feel more protective or dangerous in an academic context. See Open AI’s full policy here and these FAQs on how the company many use information shared with it.

Need help?

  • Please reach out to CNDLS if you have any questions about the use of ChatGPT in the classroom.
  • David Lipscomb, Director of the Writing Program, is available to discuss ways of thinking about the role of writing, teaching writing, and any other related issues.
  • The Academic Honor Council may also be a resource if you have questions about managing this issue with your students.