Suggestions for Teaching Writing with Generative AI

These suggestions for teaching writing with AI are built upon and extend decades of research into best practices in writing instruction, which increase the likelihood that writing in the classroom leads to engaged, deep learning and discourages the misuse of AI. Each of these suggestions are explained in more detail below.

#1: Continue teaching writing as a process and using best practices in writing instruction
#2: Develop and clearly communicate generative AI expectations and guidelines for writing assignments
#3: Experiment with integrating generative AI into different parts of the writing process
#4: When students are allowed to use generative AI for writing assignments
#5: Attend to the equitable and inclusive use of generative AI in the writing process
#6: Critique the limitations of AI text-generators as writing technologies
#7: Research how generative AI might be changing your discipline and the professional writing of your alumni and consider adjusting the types of writing you assign accordingly
#8: Explore resources & other Elon program supporting AI

#1: Continue teaching writing as a process and using best practices in writing instruction

Decades of research into best practices in writing instruction remain relevant and should continue to be central to your writing pedagogy. Focusing on writing as a process (and not only as a final product), scaffolding writing assignments, and integrating best practices not only support students in learning content and developing as writers, but they also discourage the misuse of generative AI. Continue to integrate best practices into your writing pedagogy by:

Designing meaningful writing assignments with specific audiences, purposes, and rhetorical situations and that encourage learning content and developing specific writing skills
Focusing on writing as a process by building in, for longer assignments, brainstorming, researching, writing multiple drafts, peer-response, and revising; focusing on writing
Asking students to write in a range of genres, especially those connected to specific disciplines
Incorporating different ways of writing into your classes, including informal, low-stakes writing; reflective writing; longer, high-stakes writing; and multimedia writing

#2: Develop and clearly communicate generative AI expectations and guidelines for writing assignments

Be transparent with students about your writing assignment expectations about the use of generative AI.
Add a clear “Acceptable Generative AI Use” statement to your syllabus and to individual writing assignments. Many example statements can be found online. This “AI Syllabus Language” heuristic helps you think through the stance you’d like to take with AI in your policy and provides a heuristic for developing your own AI Policy.
You might have a generic statement for your syllabus, and then more detailed statements for each of your writing assignments. Or you might have one detailed statement for your syllabus that you repeat for each writing assignment.
Your “Acceptable Generative AI Use” statement might be different for different writing assignments.
Your “Acceptable Generative AI Use” statement might range from never permitting the use of AI to permitting it in specific instances. For example class policies for classes across the curriculum and heuristics that can help you create your own policy, see “Classroom Policies for AI Generative Tools” and “AI Syllabus Language Heuristic.”
Explain your “Acceptable Generative AI Use” statement to your students and how it will apply to the writing assignment.

Explain the consequences if a student violates your “Acceptable Generative AI Use” policy.
Consider your stance, and your discipline’s approach, toward plagiarism and generative AI when developing your “Acceptable Generative AI Use Policy.” As this infographic asks, “Which of these you consider ‘cheating’? Which of these is relevant to our students’’ future? Which of these would you use in your work as an adult?” (Ditch that Textbook)

#3: Experiment with integrating generative AI into different parts of the writing process

Writing technologies are always changing, which in turn alter how we write. In the suggestions below, students are encouraged to use their disciplinary knowledge and critical thinking about their writing situation in order to use generative AI effectively. Each suggestion also requires students to consider how their prompt elicits different outputs (i.e., they engage “prompt engineering,” which is a kind of critical thinking), to critically evaluate AI output, and to engage in the “human-AI loop” (refining their prompts to ask additional questions or to submit additional prompts). Here are some types of example prompts students might ask different AIs during different parts of the writing process:

Brainstorming

Explore ideas from multiples perspectives
Overcoming writer’s block: ask for an opening into a topic or for an opening sentence
Submit background information about a situation/topic/debate, and then ask what a particular audience would find persuasive – critique the output using your understanding of the audience and then ask follow-up questions – about content, genre, medium/media.
Give the AI a “persona” (resistant, supportive) a specific demographic (age, geographic location, religious affiliation), or a specific expertise (lawyer, scientist) then submit prompts that mimic having a “conversation.” Use critical thinking and disciplinary knowledge to compose your prompts in response to output.
Ask questions to learn more about topic or audience demographics, psychographics.

Researching

Identify the top 5 or most important articles on a topic; critique the output, and ask follow-up questions, refine your focus.
Copy articles (free) or upload pdfs (paid) and then ask AI to generate:

– summaries
– synthesis statements
– comparison of methodologies, results, populations
– critique the output and revise

Specialized research AIs like Elicit (some paid)
Submit a research question, then broaden or narrow it by adding in a new focus or asking for a perspective to be removed (based on what output is returned). Continue to narrow or broaden based on your critique of the output and understanding of rhetorical situation.
Ask for interesting research on X topic or that includes or excludes X, Y, Z topics. Critique the output and then refine it by submitting follow-up prompts.
Give an example of gaps in research to “train” the AI. Then ask it to identify gaps in research for a given topic/ field; critique output, refine accordingly.

Drafting

Generate an outline and critique the output’s strengths, weaknesses. Revise the outline on your own and then write your own draft.
Compare a student-written draft with an AI generated draft; compare, critique both their strengths and weaknesses, revise accordingly.
Ask which topics are important to include for X audience, or the organization that would best appeal to Y audience. Critique the output and make topic and organization decisions accordingly.
Generate a draft of a specific section, using critical thinking back and forth prompting. Then critique its strengths and weaknesses for different genres, audiences, purposes.
Use chain-of-thought or “step” prompting to draft sections of a longer text, critiquing the output along the way and adjusting your prompts accordingly.

Revising
When revising, it’s always the writer’s job to use critical thinking and understanding of the rhetorical situation to decide what kind of feedback to ask for and which suggestions to accept/reject.

Use critical thinking and understanding of the rhetorical situation to ask for specific kinds of feedback and to decide which suggestions to use and which to reject.
Style: Is the style appropriate for X audience? Revise for X style.
Organization: Are the paragraphs organized logically, effectively for this purpose? Is each paragraph coherent, cohesive?
Clarity: Are the topic sentences clear and appropriate?
Argument: Is my thesis supported by the evidence?
Content: Does the introduction fit with my thesis, give the reader enough context to understand my argument? Is any information or important viewpoints missing?

Editing

Use critical thinking to decide what kinds of edits to make and to decide which output to accept/reject.
Grammar and Spelling: AI can identify grammar errors, spelling mistakes, punctuation issues, and suggest corrections
Clarity and Style: AI can suggest changes to sentence structure, word choices, and transitions. It can check for passive language or overuse of words.
Consistency: AI can help ensure consistency in terms of tense, tone, and terminology.
Citation and Formatting: AI can check that citations are correctly formatted according to the style guide. AI can write/format citations.

#4: When students are allowed to use generative AI for writing assignments

Teach students to use it critically, ethically, and responsibly, just like with other writing technologies (Wikipedia, search engines, Word autocorrect, etc.).
Discuss how AI outputs should never be used entirely as they are and without revision, but rather as starting points in the Human-AI Loop (content is created not just by the AI statistical probability, or just by a human, but rather by a human revising AI outputs given the rhetorical situation)
Experiment and practice using the AIs together as a class as part of the writing process
Critique the generative AI outputs together as a class. For example, in what ways is the output effective or ineffective? appropriate in content, style, or form? appeals to or fails to appeal to a specific audience?
Critique the benefits and limitations of using generative AI in the writing process
Ask students to write a reflection when they’ve used generative AI for writing assignments. For example, which AI did they use? in what ways were the outputs effective, ineffective? how did they revise the output (the Human-AI Loop)?
Ask students to submit the transcripts created while using generative AI

#5: Attend to the equitable and inclusive use of generative AI in the writing process

If you choose to integrate generative AI into your writing pedagogy, ensure that students have equal access to the technology and to training and support. If you are using an AI that requires a subscription, include the cost in your syllabus and arrange for ways for students to access the technology if the cost is prohibitive, or commit to using only free AIs. Explore how different AIs (like text-to-speech or speech-to-text) might enhance the writing process for students with different needs.

#6: Critique the limitations of AI text-generators as writing technologies

While AIs are continually emerging and changing, some common limitation include:

Lack of recent and information
Inability to reason or deeply understand context
Sensitivity to input phrasing, prompt-engineering
Potential for AI to generate plausible but incorrect information (especially with older Ais)
Bias from training data
Lack of emotional and moral understanding
Difficulty with ambiguity and context switching

#7: Research how generative AI might be changing your discipline and the professional writing of your alumni and consider adjusting the types of writing you assign accordingly

We know that some professions have been integrating generative AI into their writing process for a while, and there is much speculation about how generative AI might alter professions and lead to lob loss (for example, see “AI is on a collision course with white-collar, high-paid jobs-and with unknown impact”). Talking with students who’ve done internships and alumni from your programs can help us understand how their professional writing lives might be impacted by generative AI. Consider applying for a CWE “Elon Alumni Who Write Grant,” which will award you $300 for inviting alumni to speak on campus (in-person or zoom) about their professional lives, including the kinds of writing they do at work and the extent to which they might be using generative AI. Consider teaching how professionals use AI somewhere in your curriculum.

Some example professions that have been using generative AI include: journalism, marketing and advertising, finance and banking, data science, software development, accounting, professional writing and editing, law, education, customer services, scriptwriting, real estate, healthcare, and social media.

#8: Explore Resources & Other Elon Programs exploring AI in education

Explore the many writing pedagogy resources that are available and visit other Elon programs doing valuable work with generative AI.

The below links are being updated and will be available soon.

AI Use in Elon’s Writing Center: Guidance for Faculty and Students
CATL’s Artificial Intelligence and Your Teaching
Association of Writing Across the Curriculum’s (AWAC) Statement on Artificial Intelligence
MLA/CCCC Joint Task Force on Writing and AI

Writing
ChatGPT 4 ($20/month)
ChatGPT 3 (free)
“35 AI content generators to explore in 2023” (March 2023)

Research & Research Writing
“Top 10 tools for Academic Research” (March 2023)
Elicit: The A.I. Research Assistant
“ChatGPT for Research Paper Writing: A Prompt Guide for Teachers & Students”

Images
The best AI image generators
DreamStudio (Stable Diffusion) for customization and control of your AI images

A.I. Acceptable Use Policies
“Classroom Policies for AI Generation Tools”
“A.I. Syllabus Language Heuristic”
“My class required A.I. Here’s what I learned so far”