Dialing It In: Writing Prompts

Prompting learners in instructional settings traces back to Socrates, who asked notoriously tough questions: What is virtue? How do you define beauty? By what standards do you judge things to be good or bad? Socrates, who lived during the codification of logical argument, aimed to stimulate what some call ‘critical thinking’ long before Benjamin Bloom created a taxonomy of thinking in the 1950s.

Believing that eternal salvation belongs only to those who embrace the teachings of Jesus, in 1529 Martin Luther wrote his catechism, an easy-to-use set of questions/prompts and expected answers about the Bible which spiritual leaders could employ to give poor villagers access to a path to heaven. Luther wrote the catechism such that the basics of Christianity would be learned not by rote “the way monkeys do it,” but by understanding.

1n 19th century America, teachers began to hold spelling bees, competitive events won or lost depending on spelling skills. Teachers prompt students to produce a legal spelling of a word; students often have the right to ask for the teacher to use the word in a sentence before they pronounce the word, spell the word, and pronounce it a second time.

Interestingly enough, the first commonly mentioned piece of educational research published in the 1890s by Joseph Mayer Rice relied on a written version of a spelling bee, called a “spelling test,” as a way to determine whether the amount of time teachers devoted to spelling activities was paying dividends. His interest was in reforming schools to include art, music, and other subjects beyond the basics, and he documented that the amount of time schools spent on spelling instruction bore no relationship with spelling achievement.

Hugh Mehan, a sociologist of education at UC San Diego, studied classroom patterns of discourse ethnographically and discerned a widely used catechetical model of instruction in the 1970s. Known as the IRE model, students taking methods courses in teacher preparation programs sometimes mistakenly believe that Mehan recommends practicing it. In fact, he critiqued it, drawing attention to an impoverished method with significant implications for high poverty learners.

The model goes like this: I stands for ‘initiation,’ R stands for ‘response,’ and E stands for ‘evaluation.’ Teachers initiate interactions with students by posing problems or asking questions, students respond by giving an answer or providing information, and teachers evaluate responses, letting students know if they were right, wrong, or somewhere in between.

His analysis vivified implications for reforming classroom dynamics, teacher-student power relationships, and the ways in which knowledge is constructed and evaluated in educational settings reminiscent of Martin Luther’s critique of rote instructional methods teaching students “the way monkeys do it.” Mehan's work inspired further study of the impact of discourse structures on learning.

During the 1980s and 90s, the notion of an instructional prompt or an initiation exploded into more complex forms rooted in John Dewey’s philosophy of the activity curriculum, opening up opportunities for negotiation between teacher and learner, introducing a power-sharing dimension, positioning learners with responsibility for decision-making, making space for self-assessment, peer feedback, and collaboration. Teachers “prompt” students in problem-based learning (PBL), for example, but the responses students give are anything but rote:

“Broad problem posing: A teacher writes the question on the board: “Are organic fertilizers better than commercial fertilizers?” The question is purposively broad and requires student teams to clarify the question before even beginning to address it.”

The concept of writing prompts has a long history. In ancient civilizations and during the medieval period, writing prompts were used to teach technical and linguistic matters. Students were provided with model texts to copy, preparing them to serve as scribes. Individuals from the higher echelons of society would no more learn to write than they would take an interest in cleaning out a horse stall. Often, prompts were religious or philosophical texts, but there was no intention of supporting students in understanding their contents.

“Writing a medieval text with a quill is hard work.  The pen could only make a more or less downward movement because of how the nib was cut. It meant that letters had to be broken up into multiple pen strokes. This made writing a very slow process: a Bible could easily take a year to complete.”

Today, most teachers conceive of a writing prompt as a recipe of sorts, a description of a task or situation calling for the production of an expository or argumentative text, ranging from, say, a paragraph to an essay. Creative writing prompts are also used, particularly in the elementary grades. Since the late 1970s writing prompts accompanied by scoring rubrics have been used to both teach and assess writing “on demand,” and states from California to Kentucky to Vermont have created large-scale writing measurement systems to function as tools for program improvement as well hold schools accountable.

In my view, task-centric writing prompts are categorically different from catechetical Q and A models of teacher-student interactions. To be sure, prompts are used to present teacher questions requiring student response for evaluation much like Mehan’s IRE—we call them “essay exam questions.” Task-centric prompts, however, specify discrete demands intended to evoke open-ended responses. Guidelines for writing like topic, genre, audience and purpose direct the writer to produce a text within the parameters of prompt specifications. This modern prompt represents a completely different paradigm in contrast with the medieval prompt.

Writing instruction grounded in a task-centric prompt model lost some of its luster last year with the advent of natural language algorithms. Before AI, teachers could prompt students to write and expect them to “do the work” themselves. The work as it is presented in writing prompts, however, just happens to be tailored nicely to the algorithms AIs deploy to produce their output. Prompt-driven task-centric writing assignments and assessments are on a collision course with AI. Many teachers see limited options: require students to do their writing in class, forbid use of AI, or penalize students for cheating.

In my view, a way to transform task-centric prompts and make peace with AI isn’t beyond reach. It’s true that in order to write an essay, writers need a sense of audience, occasion, subject, and purpose. It’s also true that AI can effortlessly generate an essay when it is provided with such information, even if it’s incomplete: “Write a 750 word essay explaining how gravity affects the tides.” The audience is implied (the teacher), the purpose is implied (to get a grade), the occasion is obvious (a class assignment)—and we have a genre and form.

The principal of authenticity provides insight into how prompt theory might be expanded to provide richer opportunities to learn and to open up strategies for using AI as a research tool. In the real world writers sometimes are required to write in response to external specifications, for example, writing a personal statement for a university application. But because the writing is real, the writer is invested in the purpose, the writer thinks deeply about the audience, etc. Evaluation in reality doesn’t come in the form of a grade.

Conventional uses of task-centric writing assignments have the patina of authentic writing, but they are artificial in that students are given the parameters. They don’t have to solve the problem a writing task presents on their own, it’s been handed to them, and their motivation is completely different—arising internally in the case of real writing, arising externally in the case of a prompt. Turning to AI is not just easy, it’s tempting especially if the writing task is personally meaningless.

Before writing this essay, I had to solve the writing problem for myself. My audience is known to me in some detail in that I know roughly a third of my subscribers. Some have been in classes with me, I’ve taught with others, I’ve had email conversations with some, I get comments (I could really use more feedback from you), I get information about which posts have been opened more than once, I can glean some information from your email addresses. I believe I must be doing something right because you sometimes read the posts, and I reflect on posts that garner more attention.

[The following excerpt is drawn from a session with ChatGPT4 and is provided here as an illustration to think about.]

User: Prepare a writing assignment that requires students to embrace their own purpose in composing a text and invites them to consider optional lines of development. Provide students with task specifications like those that would apply if they were to submit the text for publication.

Smith4: Your task is to write a comprehensive essay that explores the ethical implications of artificial intelligence in contemporary society. Consider the following questions to guide your examination:

1. How do AI systems influence privacy rights, and what measures should be in place to protect individual data?

2. In what ways can AI perpetuate or exacerbate existing biases? Analyze recent examples where this has occurred and discuss potential solutions.

3. Discuss the concept of accountability in AI decision-making, particularly in high-stakes areas such as healthcare, criminal justice, and autonomous vehicles.

4. What are the implications of AI on future employment? How should society prepare for potential job displacement or transformation due to AI and automation?

5. Can AI systems be genuinely ethical, or do they simply reflect the values of their creators? How should the development of ethical AI be approached?

Your essay should include a critical review of the current literature on the ethics of AI, incorporating relevant theories, studies, and examples. Be sure to address potential counterarguments and propose thoughtful solutions based on your analysis.

Criteria for evaluation will include the depth of your research, clarity of argument, originality of thought, and the quality of your proposed solutions. Your essay should be well-structured, with a clear introduction, body, conclusion, and properly cited sources throughout.

Expectations:

- Length: 3500-5000 words

- Format: APA or MLA formatting style

- Sources: Minimum of 10 peer-reviewed scholarly articles, books, or credible online sources

- Deadline: [specific deadline, e.g., one month from the assignment date]

This assignment will require you to engage in extensive research and critical thinking. Your final essay should provide a nuanced view of the ethical landscape surrounding AI, along with practical proposals that address the challenges identified.