Created through the Digtal Pubishing Initiative at The Creative Media and Digital Culture program, with support of the OER Grants at Washington State University Vancouver.
"When a literary work interrogates the inscription technology that produces it, it mobilizes reflexive loops between its imaginative world and the material apparatus embodying that creation as a physical presence. Not all literary works make this move, of course, but even for those that do not, my claim is that the physical form of the literary artifact always affects what the words and other semiotic components mean." - N. Katherine Hayles, Writing Machines (2002)
In her book Writing Machines, N. Katherine Hayles reflects on how successive waves of inscription technologies—from quills to word processors—have fundamentally reshaped what it means to write, read, think, and ultimately to be human.
Hayles introduces the concept of "technotexts", literary works that emphasize the materiality of their medium. As artificial intelligence increasingly takes on tasks once considered uniquely human, like writing, we are challenged to confront how these technologies redefine core aspects of identity and creativity. New inscription technologies demand ongoing human experimentation to evolve new literary forms. While AI-generated technotexts present intriguing possibilities, given how easily text can be manipulated with language models, relying solely on AI to produce a conventional novel or poem—without the experimentation, creativity, and ambition of human writers—misses the deeper potential of this collaboration.
AI signals a new leap in writing technology. AI language models can generate fluent, coherent text on almost any topic in seconds, raising important questions about authorship and the human role in creating written works that shape culture. Consider a human writer with a hired assistant for research, outlining, dictating, copyediting, and even employing complex or time-consuming procedures on a text. This partnership could free the writer to focus on creativity and invention. In a sense, we already accept such collaboration in ghostwriting, where individuals hire others to write in their voice—common in fields like business, medicine, law, and fiction publishing. For instance, bestselling author James Patterson often collaborates with co-authors to produce multiple novels each year.
The idea of AI as a writing assistant encourages us to rethink the collaborative potential between humans and machines. Rather than replacing the human element, AI can enhance creative processes by helping with structure, grammar, and phrasing, while enabling writers to focus on original thought and style. In educational contexts, AI can help students discover their unique voice, allowing them to concentrate on ideas rather than mechanics. This integration has the potential to transform how writing is taught, shifting the focus from formulaic approaches to fostering deeper engagement and enjoyment of the craft. Instead of diminishing human creativity, AI expands our ability to achieve more in writing.
However, this shift requires a reassessment of how we teach writing. Much of contemporary writing instruction prioritizes rigid, formulaic methods, particularly in academic contexts, where students often adopt a voice that feels disconnected from their own. AI challenges us to rethink what writing should be. Traditionally, essay writing was a form of personal exploration, but it has become increasingly mechanical in educational settings. By incorporating AI, we can return to a more organic form of writing—one that emphasizes meaning and clarity over strict adherence to structure, encouraging students to use writing as a tool for original thinking and self-expression.
Large Language Models (LLMs), such as ChatGPT, hold tremendous potential for enhancing research and writing processes. When connected to web search capabilities and trained on specific datasets, these models can provide comprehensive and nuanced insights on a wide range of topics. For instance, when tasked with writing an essay on early film copyright cases, an LLM can potentially access and synthesize data from legal archives, historical records, and scholarly articles, delivering a well-rounded perspective. This capability to draw from vast databases can significantly expedite the research phase of writing projects, making it easier to gather relevant information quickly and efficiently.
However, utilizing LLMs for factual research presents challenges, particularly concerning the accuracy of the information provided. LLMs have a known tendency to generate fictional or inaccurate content, which poses a risk when relying on them for critical research tasks. Ensuring the veracity of facts becomes a crucial responsibility for the human user. It is essential to cross-check the information generated by the LLM against reliable sources. Specifying the need for factual and truthful responses in prompts can help mitigate some inaccuracies, but constant vigilance and verification are necessary to maintain the integrity of the research.
Despite these challenges, LLMs can be invaluable tools in probing complex topics and generating ideas. For example, initiating a research project on the influence of Renaissance art on modern visual culture with an LLM can yield a broad overview and highlight various facets of the subject that might warrant further investigation. By asking detailed questions and specifying the need for factual accuracy, researchers can draw on the the LLM’s capabilities while maintaining control over the quality of the information. Over time, as LLM technology evolves, the collaboration between human researchers and AI is likely to become more seamless, with improved mechanisms for ensuring factual accuracy.
LLMs offer fiction writers a range of possibilities for conducting research, especially when exploring specialized or unfamiliar topics. Imagine an author planning to write a contemporary novel or screenplay about deep sea diving. The author could use an LLM to assist in various aspects of their research, experimenting with different methods depending on their creative preferences. While some might find value in using the LLM to generate character names or plot suggestions, others might prefer it to function as a brainstorming partner, avoiding direct intervention in the creative imagination. The key is to tailor the use of the LLM in ways that enhance rather than constrain the writing process.
Here are several techniques a fiction writer could employ when using an LLM to research for a project on deep sea diving:
The point of using an LLM is not to replace human creativity but to explore new ways of collaborating with technology to enhance and enrich the creative process. Authors can experiment with these techniques, selecting the methods that best align with their writing style and objectives, creating a unique synergy between human imagination and machine learning.
The integration of AI into essay writing for academic settings presents a double-edged sword. On one hand, students struggling with the mechanics of constructing a coherent, grammatically correct essay with citations could use AI writing tools to get a solid draft in place - allowing them to focus more energy on ideation, research, and critical thinking. This does not mean that students should skip learning the mechanics of essay construction, but an LLM might break down essay structure, introduce various citation styles, and elements of style in grammar and syntax.
The risks of such AI tools for essay writing cannot be ignored. Lazy students may be tempted to cheat by trying to pass off AI-written essays as their own original work. This threatens to undermine the entire point of writing exercises, which is to develop critical thinking and communication skills. And like any technology, AI writing models can perpetuate societal biases and inequities if they are trained on datasets that reinforce stereotypical perspectives.
To harness AI's potential while upholding academic integrity, schools may need to adapt guidelines around proper AI tool usage for assisted writing, similar to existing policies around plagiarism, citation standards, and authorized learning aids. Rather than banning AI outright, we should explore ways to thoughtfully integrate it into the writing curriculum and leverage its strengths to become better communicators - not have it subvert the skills we aim to cultivate.
The effective use of AI, particularly large language models (LLMs), can significantly enhance the essay writing process by supporting students in various stages of their work. Below are examples of specific prompts for different stages of student writing and considerations for educators to ensure ethical and effective integration.
How it Helps: The LLM can assist in generating a targeted list of essay topics or provide a concise summary of existing research. This helps students understand the landscape of their topic and refine their focus early in the process.
How it Helps: The LLM can clarify the assignment requirements and help students outline the necessary components, ensuring they meet the educator’s expectations. This allows students to approach their work with a clear plan.
How it Helps: The LLM can offer feedback on initial ideas, suggest ways to strengthen arguments, and help anticipate counterarguments. This supports deeper critical thinking and a more robust essay.
How it Helps: The LLM can help structure the essay logically, ensuring that each section flows into the next and that all necessary points are covered. Students can then build their drafts around this outline.
How it Helps: The LLM can offer real-time feedback on drafts, suggest alternative phrasings, and help refine ideas. This allows students to focus on content while receiving support on structure and clarity.
How it Helps: The LLM can assist in polishing the language, correcting errors, and suggesting stylistic improvements. This stage emphasizes the importance of clear communication and ensures the essay meets academic standards.
How it Helps: Students should always fact-check AI-generated content. The LLM can assist in comparing facts and suggesting alternative ways to phrase ideas to maintain originality, but the final responsibility lies with the student.
By thoughtfully integrating LLMs into the essay writing process, students can enhance their writing skills while still prioritizing the development of critical thinking, creativity, and personal expression.
While AI writing assistants can be valuable aids for analytical and expository writing, they also open up novel frontiers for creative writing. The unique "logic" that underlies how AI language models generate text, can be used for procedures and constraints that can be the starting point for crafting inventive, experimental works.
Take, for example, the constraints of poetic forms like the Shakespearean sonnet - its strict iambic pentameter structure, 14-line length, and defined rhyme scheme. Feeding these specific parameters to an AI model allows it to generate new sonnets that adhere to the traditional guidelines while exploring utterly original themes, wordplay, and imagery that a human writer may never have conceived.
Similarly, AI can be a potent tool for reviving Dadaist, Surrealist, and Oulipo literary techniques designed to undermine logical conventions. For the classic "cut-up" method popularized by William S. Burroughs, an AI could blend and remix passages from wildly disparate sources into startling new narratives and ideas. Or for Oulipo constraints like N+7, where every noun gets systematically replaced, the AI's proficiency in coherent language generation means it can produce strangely coherent-yet-absurd paragraphs primed for further manipulation.
Rather than eliminating the role of the human author, these constraints and procedural techniques position writers as curators and sculptors - providing the raw clay for them to mold into refined creative works. The machine offers up an "exquisite corpse" of semi-coherent, semi-nonsensical gibberish, and the writer then rehabilitates it into something transformative through their authorial voice, selection, and compositional talents.
Mark Amerika is a pioneering digital artist whose work explores the intersections of art, technology, and narrative, often incorporating artificial intelligence into his creative process. Blurring the boundaries between author, artist, and machine, Amerika uses AI to investigate the evolving role of the artist in a digitally mediated world. His works, such as Remix the Book and GRAMMATRON, are experimental and interdisciplinary, merging elements of literature, visual art, and electronic media. Through his exploration of AI, he challenges traditional notions of authorship, agency, and creativity, positioning AI as a collaborator in the production of art and storytelling.
Sasha Stiles is a groundbreaking digital poet and artist whose work merges human creativity with artificial intelligence to explore the evolving nature of language, authorship, and consciousness. Using AI as both a tool and collaborator, Stiles creates poetry and digital art that challenge traditional literary forms, experimenting with machine-generated text and neural networks to craft pieces that feel both futuristic and deeply personal. Her works, such as Technelegy, explore the intersections of technology and humanity, posing questions about how AI can reshape the way we think about communication, creativity, and the boundaries between human and machine intelligence. Through her innovative practice, Stiles opens new frontiers in AI-assisted creative writing and digital expression.
Allison Parrish,is a poet, programmer, and professor whose work focuses on computational creativity, specifically in the realms of language and writing. She uses machine learning algorithms and natural language processing to generate experimental poetry, exploring how AI can reinterpret language and transform traditional writing practices. Parrish's projects, such as Everyword and her various text-generating bots, highlight the creative potential of AI to deconstruct and reimagine language, pushing the boundaries of poetry and authorship in the digital age.
David Jhave Johnston is a digital poet and artist who explores the intersections of AI, language, and immersive media, creating works that merge poetry with advanced computational techniques. Known for his groundbreaking project ReRites, Johnston used neural networks to generate poetry, positioning AI as a co-author and collaborator in the writing process. His work is deeply concerned with the nature of language, exploring how machine-generated texts challenge human creativity, meaning-making, and the poetic form. Johnston’s interdisciplinary practice spans interactive installations, multimedia, and computational writing, often questioning the boundaries between human and machine creativity. Through his innovative use of AI, he invites audiences to rethink the role of technology in shaping future literary and artistic landscapes.
Beyond individual creative writing pursuits, AI can catalyze new forms of collaborative human-machine writing. One intriguing example is the AI-augmented revival of the Exquisite Corpse technique, the Surrealist parlor game where collaborators take turns writing sections of a story without seeing the full preceding text.
David Jhave Johnston's "Rerites" is a compelling example of human-machine collaborative poetry. This project involves Johnston working in tandem with AI algorithms to generate poetic texts. The collaboration unfolds through an iterative process where Johnston feeds the AI model with a vast corpus of poetry and prose, which the machine then uses to generate new lines of text.
Johnston's role is not merely that of a passive recipient of the AI's output; instead, he actively engages with the machine-generated text, curating and editing the lines to shape them into cohesive poems. This dynamic interplay between human creativity and machine computation results in a unique blend of styles and themes, showcasing the strengths of both human intuition and AI's capacity for vast data processing.
This sort of human-AI collaborative writing dynamic has already begun emerging in experimental interactive fiction projects and even mainstream video games with procedurally generated storytelling elements. But we are only scratching the surface of what is possible as language models grow more sophisticated and writers become adept at choreographing the machine's strengths to amplify their own creative visions.
To culminate our exploration of AI for creative writing, this unit's exercise will have you actively experiment with different AI collaboration and generation techniques. Working individually or in small groups, you will use an AI language model (such as GPT-3, custom models you've trained, or other available tools) to produce an original short story, poem, or other creative writing work.
Beyond just leaning on AI for generating drafts, the goal is to fully embrace the unexpected detours and surprising elements that emerge from interrogating the model's outputs. Don't be afraid to follow the strange rabbit holes the AI takes you down - that is where you'll discover the most innovative and imaginative ideas to shape into your finalized piece.
This exercise is designed to help you explore a subject you're unfamiliar with and reflect on the process of learning about it with the assistance of an AI. The final product will be a personal essay that captures both your new understanding of the topic and your experience of inquiry. Follow these steps to complete the exercise:
Ultimately, this exercise aims to have you transcend just using AI as a utility and explore how cooperative human-machine orchestration can spark new forms of creative and intellectual expression. As AI's role in writing continues evolving, developing these skills for fruitful collaboration will be crucial for forging novel storytelling and non-fiction possibilities.