Composing with AI – Research & Scholarship

Traditionally, the writing process is conceptualized to be (1) a series of recursive steps (aka stages or intellectual strategies); (2) a cognitive, problem-solving activity; and (3) a creative, intuitive, organic, dialogic process that writers manage by listening to their inner speech and felt sense. These theories of composing form the foundation of composition studies, rhetoric, an computers and writing -- three disciplines, three methodological communities, who investigate ways new technologies -- e.g., the pencil, the printing press, radio, TV, internet, AI -- impinge on composing processes. The article below contributes to this scholarly conversation. Specifically, it analyzes how writers in school and workplace contexts are composing with AI, whether that the meaning they are composing is expressed in digitial media, painting, photography or other media. Specifically, this article analyzes the constraints and affordances of GAI (generative artifical intelligence) tools from the interpretive lens of writing studies. It provides a summary of research on the topic, using multiple, mixed methods, such as protocol analysis -- asking writers to vocalize their thoughts, actions, intentions, and inner speech when composing. In writing studies.

Table of Contents

Dear Colleagues, I ask that you ignore this page for a while cuz I’m just drafting here. Right now, these are just notes.

Table of Contents

Related Concepts

Composing; Composition Studies; Rhetoric & Apparatus Theory; The Writing Process

Research on AI?

This article reviews scholarship and research on these questions:

How do writers compose with AI systems, such as a, b, c. analyzes
How are Authors Using AI Tools to Write, Design & Publish their Works?
What are the the affordances and constraints of these tools on composing processes?

What Are Traditional Models of Composing?

In the past there have been three major ways of conceptualizing the writing process:

Writing in a problem-solving process.
Writing is a process of listening to your inner speech, of translating thought into language, of listening to your felt sense in order to create, communicate, and share
Writing is a process of working through recursive stages (aka, strategies, intellectual activities, steps)

Writing as a Problem-Solving Process

According to Linda Flower and John Hayes (1977) writing should be thought of as a “thinking problem,” a “problem-solving process,” or “cognitive problem solving process.” In this model, a writer’s aim of discourse directs all writing processes. From this interpretive framework, the writing process is largely cognitive and guided by the author’s intention. Here the exigency, the rhetorical situation, is conceptualized as a Task Environment. This initial model didn’t address research and collaborationa;

cognitiveprocess - Writing Commons — Source Hayes J R Flower L 1980 Identifying the Organization of Writing Processes In L W Gregg E R Steinberg Eds Cognitive Processes in Writing An Interdisciplinary Approach pp 3 30 Hillsdale NJ Lawrence Erlbaum

It as a hierarchical set of subproblems arranged under a goal or set of goals. The process then is an iterative one. For each subproblem along the way — whether it is making a logical connection between hazy ideas, or finding a persuasive tone — the writer may draw on a whole repertoire of procedures and heuristics” (Flower & Hayes, 1977, p. 460-461).

Writing as a Process of Listening to Your Inner Speech & Inner Voice

Van Gogh's 'Starry Night' — a masterpiece born from a mind that saw the world differently. Just as art benefits from diverse perspectives, educators must appreciate neurodivergent students for their unique and valuable ways of interpreting the world and composing.

Historically, especially since the Romantic Poets, the writing process has been conceived of as a creative, intuitive, organic, dialogic process that writers manage by listening to their inner speech and following their felt sense. This is the process of “creatives” — people who don’t focus during prewriting on rhetorical analysis, which typically drives the composing process of problem-solving approaches to writing.

From a scholarly framework, this composing model is grounded in the work of Vygotsky and his theory of “inner speech.”

Writing as a Recursive Series of Steps or Stages

writing-steps - Writing Commons — Photo Credit Moxley

Since the 1960s, the writing process has been defined to be a series of steps, stages, or strategies. Most simply, the writing process is conceptualized as four major steps: prewriting, drafting, revising, editing. That model works really well for many occasions. Yet sometimes you’ll face really challenging writing tasks that will force you to engage in additional steps, including, prewriting, inventing, drafting, collaborating, researching, planning, organizing, designing, rereading, revising, editing, proofreading, sharing or publishing. Expand your composing repertoire — your ability to respond with authority, clarity, and persuasiveness — by learning about the dispositions and strategies of successful, professional writers.

Composing with AI – A New Model of Composing

An illustration of a college student looking at her graded paper online. The paper has earned C+ and this comment, "You are entitled to your opinion. But you are not entitled to your own facts." — opinions are not facts

is it plagiarism when writers dialog with AI tools — sometimes one tool but often multiple tools — on a sentence-by-sentence or paragraph basis? Is it plagiarism when writers ask Semantic Scholar, Consensus, or Consensus to summarize research on a topic, perhaps writing annotated bibliographies. After all, in the real world writers are using multiple tools to refine their thinking, just as in the past students might have visited the writing studio, peer-reviewed drafts with peers, or received critique during the writing process from teachers. When composing with AI, writers may upload research and scholarly studies their preliminary search identified as canonical to ChatPDF to more quickly identify scholarly conversations and knowledge gaps. Next, they may use LitMaps or ResearchRabbit to visualize citation relationships and knowledge claims among citations. And then based on what they read they may begin composing from scratch or they may drop snippets from the tools they used into another tool like Perplexity and ask it to write a first draft. They may then take snippets from that — perhaps just a sentence or two — and then ask ChatGPT to expound on it. Then they may take that result or just a fragment of that and ask Claude to help with revision, editing, or proofreading. Given this milieu, the MLA and APA citation guidelines are out of touch. Their guidelines are as helpful as saying to NASA astronauts they are expected to walk on the moon without setting foot on the moon. In response to these new writing processes, academe is in turmoil. This indeed is a paradigm shift in writing studies, forced by outsiders. Lance Eaton’s compilation of course syllabi polices demonstrates that some teachers (especially in the humanities) consider any use of AI tools to be unethical — an act of plagiarism. Some teachers — especially those who reject the fair use argument that AI leaders such as OpenAI and Microsoft make to empower them to scrape the internet and copyrighted works to build their LLMs (Large Learning Models) — believe any use of AI violates academic integrity conventions.