Teaching Writing with Generative AI

The KNIGHTLYnews is an online forum where FWS instructors and other teachers of writing can swap and share ideas for best classroom practice. Weekly posts are designed to help teachers develop lesson plans and writing assignments, and respond to classroom challenges by introducing new teaching tools and sharing emerging pedagogical ideas. Posts also direct readers to program and campus resources that support teaching and learning, and provide opportunities for peer collaboration and mentorship. #teachlikeabear

I have always been an early adopter when it comes to technology in the writing classroom, so when news about ChatGPT dropped several semesters ago, I began exploring with students this digital writing tool to consider if and how it could support and enhance their writing processes. Eventually, I also welcomed it as a useful teaching tool. 

 

My FWS, titled “Food for Thought,” features ongoing practice in slow learning and close reading as students work with peers to draft and revise substantial writing projects. Learning how to write strong summaries is a cornerstone. In the process of writing, revising, and discussing summaries, students have multiple occasions to develop their reading and writing process skills (writing to learn) and for us to name and talk about academic writing conventions (learning to write). 

 

Here is a three-part activity sequence from my FWS that illustrates how I embedded GAI in ways that reinforced and reinvigorated existing classroom practice. 

 

Summary Writing | Sequence of Activities

Materials 

  • Chat GPT 3.5 (free)
  • 3 articles widely available on the internet (NOT behind a paywall)

 

Activities

  • 3 In-Class Summary Writing Assignments
  • Nested within a sequence of three assignments that expect increasing proficiency in writing summaries: Academic Argument, Annotated Bibliography, and Research Paper

 

Learning Goals

  • Use GAI and summary writing to learn about writing (not teach how to write with GAI)
  • Use GAI and summary writing to reach toward course learning goals:
    • Reading → reading comprehension (what text is about & how it works)
    • Writing Process → drafting and revision
    • Writing → discourse conventions, focus, clarity, concision, voice, and style
    • Collaboration → collaborative writing and peer review
    • Research → identifying, accessing, and using information
    • Metacognition → writing lexicon and strategies for evaluating writing
    • Critical Thinking → using writing, the writing process, and writing tools to clarify and deepen thinking

 

Activity 1 | WRITE a summary of Article 1 & COMPARE with AI-generated summary

 

What we did

After a class discussion about Michael Pollan’s essay, “Escape the Western Diet,” students composed 50-word summaries.

  • We added summaries anonymously to a shared GoogleDoc (mine was generated by ChatGPT)
  • Students read all of the summaries
    • Voted on the strongest one
    • Created a rubric of essential features for summaries
    • Revised summary drafts with peers

 

What we Learned

  • Most students suspected that my summary was AI-generated, noting that it:
    • captured a broader and less subjective reading of the text and
    • did not include many terms and phrases from the text. 

 

Reflection

While student's findings about generative AI were not terribly noteworthy, especially for those who were already familiar with ChaptGPT, the activity’s broader learning outcomes were significant: The guessing game built student confidence – there was no clear winner! – and the discussion raised important questions about purpose, tone, and voice.

 

Activity 2 | REVISE an AI-generated summary of Article 2

 

What we Did

  • After a class discussion about Mary Maxfield’s essay, “Food as Thought: Resisting the Moralization of Food,” students read a 100-word summary generated by ChatGPT.
  • Students worked in pairs to revise the summary and posted revisions in a shared GoogleDoc.

 

What we Learned

  • Students read all of the summaries and worked with partners to name revisions:
    • corrected content errors
    • added more specific details
    • made summary more concise
      • omitting repetition
      • combining sentences to make syntax more complex
    • tended to word choice
      • to capture tone
      • to honor key terms and phrases in original text

 

Reflection

In this activity, our learning about generative AI became significantly more robust and so too did their ability to understand and talk about writing, especially its building blocks at the sentence level. Students were bolder with their critiques of the ChatGPT draft and moved more swiftly from critique to application in their revisions. 

 

Activity 3 | Use generative AI to CREATE a summary of Article 3

 

What we Did

  • After a class discussion about Lisa Miller’s Newsweek article, “What Food Says about Class in America,” students used ChatGPT to compose a 300-word summary.
  • They were instructed to keep track of their prompt engineering – their initial prompt and any follow-up prompts – to create a summary that “is as close as possible to one you might actually write!”
  • When students paired up to share their processes:
    • Several struggled to even get started (signing up, composing a first prompt).
    • Many found the prompting process challenging, even after demonstrations in previous activities.
    • Students achieved more satisfactory overall results when their follow-up prompts excerpted short passages from the text.

 

What we Learned

  • Students shared results with their partners, noting that ChatGPT:
    • was even more repetitive when prompted to compose longer summaries
    • needed to be prompted to include essential details, and then often made them up or mischaracterized them
    • relied on common templates when “reading” the text and generating outputs – sometimes looking in the wrong places for important information and poorly capturing and ranking key ideas
    • generated nearly identical summaries for every member of the class

 

Reflection

While students noted results similar to what they observed in activity 2, here, students were more struck by the fragile foundations upon which ChatGPT constructed its responses. Students responded differently to the realization that ChatGPT made use of common templates. More experienced writers regarded this discovery as a challenge – to reach beyond the template. Less experienced writers, though, seemed grateful for a differently powerful resource to facilitate their transition to academic genres and other writing conventions.

 

What I Learned about GAI & Teaching Writing

 

Embedding GAI elements into typical summary writing activities:

  • brought unique energy to classroom work  – Could students detect the bot? Could they beat the bot? Could they improve on the bot? Could they maneuver the bot?
  • deepened learning about course readings and writing strategies in ways that required students to write and respond to writing.
  • A bonus feature of this assignment sequence is that students also began developing critical AI proficiency.

 

What I Learned about GAI & Students

 

  • Students need practical support learning AI tools –
    • from basic support (what it is and how it works)
    • to critical engagement (how to use it strategically and responsibly)
  • Students need guidance on
    • when they can use it (course and workplace policies vary widely) and
    • how to find out rules and expectations (asking is awkward)

 

Reflections

 

As I have demonstrated here with these 3 representative classroom activities, GAI was easily integrated into my classroom practice in ways that fully supported, even enhanced my broad instructional goals:

  • increase access to learning opportunities where students can safely practice and play with language and ideas 

  • prepare students to 

    • use writing to explore ideas, deepen thinking, and develop scholarly voices

    • ask critical questions about the ways they express themselves, make meaning, and participate

  • promote writing as a collaborative process whether that collaboration is human or artificial. 

 

Links to resources

 

More news

View all news
KNIGHTLYnews
Top