Why Teach Writing in an AI World?
SEPTEMBER 16, 2024
And what, you ask, does writing teach us? First and foremost, it reminds us that we are alive and that it is a gift and a privilege, not a right. —Ray Bradbury, Zen and the Art of Writing
Why do we write? And why do we teach writing? In a world where Large Language Models (LLMs) are now ubiquitous, these questions have taken on a new urgency for students and teachers—and, indeed, for professional writers. Writing instruction in the AI era must focus on helping students discover and express their own unique ideas, voice, and purpose.
Why Do We Make Students Write?
Writing instruction has long been a staple in K-12 instruction. It is famously and colloquially known as one of the “three Rs” of grammar school: “reading, ‘riting, and ‘rithmetic.” As teachers, we use writing both as a means to assess our students’ knowledge and thinking and as an end in itself. The ability to express oneself clearly in writing, using proper grammar and mechanics, is considered by many to be one of the essential hallmarks of an educated mind. For this reason, writing skills have historically been used as a gatekeeper for higher education and many types of employment.
How have Copilot, ChatGPT, and their LLM cousins changed the equation? There are many who seem to think that writing skills are irrelevant in a world where you can enter a simple one-sentence prompt and generate an entire essay in seconds. Instead of teaching writing, they argue, we should simply be teaching “prompt engineering” to prepare our students for the future. Are they right?
AI is very good at generating a lot of words in a very short time. An LLM can respond to virtually any prompt with a series of words that is grammatically correct (if not always factually correct) and relevant to the question asked. Sometimes, with the right prompting, it can even be funny and engaging.
However, this is missing the point of writing. We do not teach writing because we need more words in the world. We teach writing because we need more ideas in the world.
LLMs, by definition, do not generate new ideas of their own. Rather, the output they produce is generated statistically, based on all the human-generated content they have ingested. In many ways, you can think of the answer an LLM gives on a given topic as a statistical average of what other people on the internet have already said. That’s why LLM answers tend to be bland, formulaic, and repetitive. If you give your class an essay prompt, and every student puts the prompt into ChatGPT, the answers won’t be identical, but they will tend to demonstrate a lot of similarities in structure and ideas. And you will be very bored reading them.
Writing is much more than simply stringing grammatical sentences together. It is a way of exploring, expressing, and developing original ideas. Student writing is important because it cements learning, builds critical thinking skills, and provides an outlet for self-expression. The most important part of writing is the thinking that it reveals.
In a powerful Op-Ed for the Washington Post, columnist Alexandra Petri puts it bluntly:
All of the buffoons excited by the prospect of AI taking over all our writing — report summaries, data surveys, children’s letters, all tossed into the same pile indiscriminately — are missing the point in a spectacular manner. Do you know what writing is?
It is thinking in a form that you can share with other people. It is a method for taking thoughts and images and stories out of your brain and putting them into someone else’s brain. E.M. Forster quotes a woman saying, “How can I tell what I think until I see what I say?” To take away the ability to write for yourself is to take away the ability to think for yourself.
Teaching Writing in the AI Era
AI has not made writing obsolete, but teaching writing in the AI era does present new challenges and opportunities. There is no doubt that prompt engineering, AI co-writing and other such skills will be part of our students’ reality. There is no putting the AI genie back in the bottle: we know students are using AI in a variety of ways, both sanctioned and unsanctioned.
What can teachers do in the face of widespread AI cheating and overreliance on LLMs? We’ve got a few ideas.
- Get creative with assignments. While writing is not dead, the traditional five-paragraph essay almost certainly is. AI has gotten very good at answering formulaic essay prompts such as, “How did Mr. Darcy’s behavior change over the course of Pride and Prejudice?” To keep students engaged in the writing process, teachers will have to give them prompts that are more creative, meaningful, and personalized. For example, you might ask students to tell personal stories, make connections to their own lives, or respond to in-class activities and discussions.
- Give writing assignments with purpose. One way to discourage AI use is to make sure that students understand why you are asking them to complete an assignment, what they are meant to learn, and how the skills they are developing will benefit them in the future. Is the writing assignment meant to showcase their writing skills, demonstrate their understanding of a topic, encourage creative thinking and expression, or something else?
- Emphasize thinking over mechanics. Unless the express purpose of a writing assignment is to assess students’ grammar and spelling skills, the thinking students demonstrate is much more important than the mechanics. We will all have AI editing tools at our disposal to help polish our drafts and correct our mistakes (though these tools can be prone to errors themselves). What we do not want students to do is outsource their thinking to AI. Make sure students know that you are looking for their own ideas, not a summary of other people’s thoughts filtered through an LLM. Look for and reward original thinking.
- Create a safe space for experimentation. One of the primary reasons students turn to AI when faced with a writing assignment is the fear of failure. While some students may just be trying to avoid effort, for many, the desire to get the “right” answer and turn in an error-free draft makes AI a tempting tool. This becomes self-reinforcing as more people use AI in the classroom and beyond. The smooth, bland, and grammatically perfect writing turned out by LLMs has become what many students think writing “should” look like. Give students a safe space to experiment with voice and writing style and reward effort and growth rather than perfection.
- Take pre-writing offline. Some people advocate using AI tools like ChatGPT to generate an outline that students can then fill in. This is exactly backwards. Again, the most important part of the writing process is the thinking. While AI can be a useful brainstorming tool in some circumstances, ultimately, you want students to come up with and develop their own ideas. Moving more of the writing process offline can help. Once students have gathered their research, have them at least start the outlining and drafting process offline. (Thinking Maps are an excellent tool for this.) For some types of assignments, it may be preferable to have students complete the entire draft offline during class time.
- Teach students to use AI responsibly. There are responsible uses of AI tools in writing. As noted above, AI can be useful for brainstorming ideas. During the research phase, for example, students may turn to an LLM to summarize and explain technical sources they are considering citing. AI is also handy during the writing process—for example, by suggesting alternate ways to phrase something or pointing out grammatical mistakes. Students should learn how to use AI editing tools effectively, which includes understanding their pitfalls and using their own judgment in deciding which suggestions to accept. Teachers should have a clear AI policy for their classroom and tell students which uses of AI are acceptable for a given assignment.
- Don’t rely on AI detectors. Last but not least, teachers should be mindful of their own use of AI in the classroom. While there are many tools out there promising to detect AI-produced content, none of them have been proven to be reliable. Savvy students will find it easy to fool these detectors by making some small changes to a mostly AI-written piece and perhaps introducing a few strategic errors. On the other hand, these tools are known for producing many false positives—and students who are highly proficient writers are the most likely to be falsely accused. Teachers can detect AI more reliably themselves by looking for hallmarks such as duplicate structure and ideas across many essays or a writing sample that is significantly different in style and skill level than a student’s prior work. Also, be cautious of using AI as a grading assistant, as these tools tend to give very generic responses that do not always accurately reflect the strengths and weaknesses of a specific piece of writing.
Writing with Thinking Maps
Want to keep the focus on thinking in the writing process? Thinking Maps help students develop their ideas and plan their writing. Rather than turning to AI for help with outlining, teach students to develop their own thinking with Thinking Maps. Student-created Maps also give teachers visibility into a student’s thinking and writing process.
Students can use Thinking Maps to:
- Take notes and organize sources during the research process.
- Organize and develop their ideas.
- Plan their writing and map out the structure of their final written product.
Students who are proficient with Thinking Maps learn how to activate their own thinking processes and take information back off their Maps to produce a written product. Our Write from the Beginning…and Beyond training series equips teachers with the strategies they need to develop strong thinkers and writers.
Every student has something unique and original to say. Helping them discover that, and giving them the tools to do so, is a gift. The best way to combat the allure of instant AI answers is to help students discover the power of their own voices and ideas.
Contact your Thinking Maps representative to learn more about Thinking Maps and Write from the Beginning…and Beyond.
No AI tools were used in the creation of this blog.
AI and Writing: Additional Resources
More from Thinking Maps:
- AI Part I: Don’t Let Students Outsource Critical Thinking to AI
- What Makes Writing So Hard?
- The Write Stuff: Writing Across the Curriculum
- Webinar: Building the Foundations for Effective Writing
Around the Web:
- NAMLE: Youth Perspectives on AI in the Classroom
- NYTimes: What Students Are Saying About Learning to Write in the Age of AI
- NCTE: How to Keep Students Writing in the Age of AI Tools
- Leon Furze: Teaching Writing in the Age of AI: Reinventing English
- Mark Watkins: The Enduring Role of Writing in an AI Era
Continue Reading
August 16, 2024
Artificial intelligence (AI) models like ChatGPT and Gemini are profoundly changing the way people interact with technology, information, and each other. Are they also changing the way we think? We must help our students avoid the temptation of outsourcing their critical thinking to AI.
June 17, 2024
To understand contemporary issues and participate fully in civic life, students need a solid grounding not only in basic facts, but also in essential critical thinking skills. Thinking Maps can help students develop the thinking skills they need to ask relevant questions, detect bias and misinformation, connect past and current events, and understand the changing world around them.
May 16, 2024
Mastering Science Concepts and Content in K12 | Thinking Maps Support student mastery of the Core Ideas and Crosscutting Concepts in the Next Generation Science Standards (NGSS) with Thinking Maps. Learn more on the blog:
April 15, 2024
Scientific thinking empowers students to ask good questions about the world around them, become flexible and adaptable problem solvers, and engage in effective decision making in a variety of domains. Thinking Maps can help teachers nurture a scientific mindset in students and support mastery of important STEM skills and content.