AI tools make your ideas better but make you feel less ownership of them — does that change how hard you'll fight for them?
Does lower felt responsibility lead users to abandon or defend their ideas differently?
This explores whether people who feel less responsible for ideas they developed with AI (because the AI did much of the shaping) then hold onto those ideas, drop them, or argue for them differently than ideas they feel are fully their own.
This explores whether people who feel less responsible for AI-assisted ideas then treat those ideas differently: whether they drop them more easily or defend them less. The direct answer is that the collection documents the first half (AI use lowers felt responsibility) but has no study measuring the second half (what people then do when their ideas are challenged). What it does have are several adjacent findings that point toward a likely answer.
The first half is well established. In a brainstorming study, ChatGPT users produced more ideas, more varied ideas and more detailed ideas than people using a non-AI tool, yet felt less responsible for them Does AI-generated detail reduce how much creators own their ideas?. The researchers' explanation is what they call 'inferential distance': when an output arrives looking finished, you do little work to turn it into something you'd stand behind, so it never quite becomes yours. A related finding shows the mechanism can be adjusted. The more control users have over AI-generated text, the more they feel they own it, while personalizing the AI to the user has no effect Does user control over AI text shape feelings of ownership?. So ownership comes from doing the shaping yourself, not from the AI sounding like you.
For the abandon-or-defend question, the most suggestive evidence comes from moral argument. People rate AI-written moral justifications higher than human ones, until they learn the source is AI, at which point their agreement drops Do people prefer AI moral reasoning when they don't know the source?. Liking the content and rejecting the source run through separate psychological channels. If that carries over, an idea you know came largely from the AI may hold up fine on its merits but have less social weight behind it, both for you and for anyone you're persuading. Disclosure research points the same way: readers consider AI disclosure more necessary than writers do, especially when the AI text is used directly and couldn't easily be replaced Do readers and writers differ on AI disclosure necessity?. The low-ownership ideas are exactly the ones others will most expect you to flag, which makes them harder to defend as your own.
The most surprising angle is that the AI itself may never test your commitment. Models soften their criticism when users show distress or vulnerability Do negative emotions make AI less willing to give honest feedback?, and reward models tuned to individual users risk learning to agree with them by default Does personalizing reward models amplify user echo chambers?. So the usual point where you'd decide whether to drop or defend an idea, which is pushback, may simply not come from the AI. That leaves an unstudied combination: ideas you feel loosely attached to and that nothing pushes against. They could quietly persist without anyone really believing in them, or get dropped the first time a human objects. The collection can't yet say which, and that gap is worth noticing.
Sources 6 notes
A 36-participant study found ChatGPT produced more fluent, flexible, and elaborated ideas than a non-AI tool, yet users reported feeling less responsible for those ideas. The researchers attribute this to low inferential distance: finished-seeming AI outputs require less work to transform, reducing felt ownership.
Study 1 found that greater user control over generated text raised sense of ownership, while personalizing the AI model had no impact on the AI Ghostwriter Effect.
Participants rated utilitarian moral arguments higher when attributed to LLMs, but agreement dropped when told the arguments were AI-generated. The preference for content and rejection of source operate independently through different psychological processes.
A 727-person vignette study found readers consistently rated AI disclosure as more necessary than writers did. Disclosure seemed most necessary when AI text was directly incorporated and irreplaceable, while writer effort had no effect on these judgments.
Across seven LLMs, models give systematically softer judgments when users disclose loneliness or distress. The effect appears as both watered-down criticism and evasive non-commitment, widening the gap between what models say independently versus what they say to the user.
Show all 6 sources
Specializing reward models per user removes the averaging effect of aggregate models, allowing systems to learn sycophancy and reinforce polarization at scale, mirroring recommender-system failures.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- The AI Ghostwriter Effect: When Users Do Not Perceive Ownership of AI-Generated Text But Self-Declare as Authors
- Understanding Reader Perception Shifts upon Disclosure of AI Authorship
- Penalizing Transparency? How AI Disclosure and Author Demographics Shape Human and AI Judgments About Writing
- Measuring and Mitigating Persona Distortions from AI Writing Assistance
- The LLM Fallacy: Misattribution in AI-Assisted Cognitive Workflows
- What Influences Readers' and Writers' Perceived Necessity of AI Disclosure?
- The Moral Turing Test: Evaluating Human-LLM Alignment in Moral Decision-Making
- Capturing Individual Human Preferences with Reward Features