Writing with AI can look deceptively simple, but new research from Iowa State University finds that effective AI-assisted writing demands more thought from the writer, not less. The study, published in the journal Computers and Composition, followed 38 undergraduate students over two semesters and identified three threshold concepts that writers must grasp before AI becomes a reliable partner. For professional writers navigating the growing presence of AI tools, the findings challenge the common assumption that AI is a shortcut.
"Writing with AI doesn't take the work out of writing," said Abram Anders, associate professor of English at Iowa State and the study's co-author. "It changes it." He and co-author Emily Dux Speltz found that students often entered the experimental course expecting AI to function as a shortcut. Instead, they discovered that AI-assisted writing forced them to engage more deeply with idea formation, judgment, and quality control.
Writing with AI is a trial-and-error craft
The first threshold concept the researchers identified is that writing with AI is inherently experimental. Students who initially treated AI like a search engine - typing a vague prompt and accepting whatever output appeared - eventually learned that effective prompting requires planning, clarity, and the willingness to iterate. "AI isn't going to provide a 'perfect' answer or automatically spit out what you need," Anders said. "It requires trial and error - trying, testing, revising and trying again."
Fluency can hide hollow prose
The second concept addresses what Anders calls the fluency trap. AI-generated text often reads with confidence and polish, which can trick writers into trusting it, even when the content is wrong, shallow, or missing the point. "AI writes in confident sentences, uses the right tone and sounds smart," Anders said. "But that polish can trick students into trusting it, even when it's wrong, shallow or missing the point entirely." Writers who learn to interrogate AI output - checking claims, refining logic, and aligning the text with discipline-specific expectations - move beyond surface fluency to real understanding.
The writer must own the purpose
The third concept shifts the focus to human agency. Anders and Dux Speltz found that once students recognized that AI cannot generate purpose, they stopped outsourcing work and started orchestrating it. "Generative AI can't decide what it's arguing, what matters or why the writing exists," Anders said. "It's a tool that requires human direction, judgement and boundaries." This realization transformed how students used AI: rather than avoiding the cognitive load of writing, they began using it to explore possibilities, test ideas, and strengthen their arguments.
Why this matters for writers
The study's threshold concepts transfer directly to professional writing. Copywriters, journalists, technical writers, and content creators who adopt AI risk falling into the same fluency trap and shortcut mindset. Learning to experiment with prompts, treat AI output as raw material that needs critical review, and maintain ownership of purpose will separate writers who produce generic text from those whose work stands up under scrutiny. "When students learn to direct AI rather than depend on it, they become stronger writers," Anders said. "That's the skill that will matter long after the tools change."
Your membership also unlocks:
