More Than Words: How to Think About Writing in the Age of AI
by John Warner
A veteran writing teacher's argument that writing is thinking and feeling, not text production, and that much school and workplace prose was already hollow before ChatGPT. Warner supplies the mental model - writing is a thinking process whose output is a by-product - that lets a learner reason about any future tool rather than memorising this year's tactics. The revised paperback adds concrete practices for writers and teachers.
This link may earn us a small commission at no extra cost to you. Affiliate disclosure
More resources on Writing with AI
How do I cite generative AI in MLA style? (Updated and Revised)
The Modern Language Association's own rules for attributing AI-generated material: never credit the tool as author, name the model and version, describe the prompt, date the generation. Five worked examples plus guidance on chasing down sources a model cites.
AI Suggestions Homogenize Writing Toward Western Styles and Diminish Cultural Nuances
A controlled experiment with 118 writers in India and the United States found that autocomplete suggestions pushed non-Western participants toward Western phrasing and stripped cultural detail, while delivering them smaller speed gains. Measured evidence of how assistance flattens voice.
ChatGPT is bullshit
Three philosophers argue that calling model errors 'hallucinations' misleads users, because these systems are indifferent to truth rather than mistaken about it. Gives a writer a working account of why every AI-supplied fact and citation needs independent checking.
One Useful Thing
Mollick's newsletter reporting his own and Wharton's controlled experiments with language models. Posts cover prompting that survives testing, where models fail, and drafting and editing workflows, revised as models change rather than frozen at publication. AI moves fast enough that a topic with only books goes stale; this is the one continuously-updated source with real evidence behind it.
Co-Intelligence: Living and Working with AI
Four working principles for treating a language model as a collaborator: invite it to every task, stay the human in the loop, tell it what kind of persona to adopt, and assume today's model is the worst you will ever use. It teaches a durable collaboration model rather than tool tactics that expire in a quarter.