When writers call a piece authentically theirs, they mean how it was made, not how it reads, and readers usually can't see that.
What internal states do writers identify as core to their authenticity?
This explores what writers point to inside themselves (their process, experience and sense of identity) when they say a piece of writing is authentically theirs, especially now that AI can produce the finished text.
This explores what writers locate *inside* themselves, rather than on the page, when they call their work authentic. The most direct evidence in the corpus comes from interviews with 19 professional writers who co-write with AI. They didn't define authenticity by the finished text. They placed it in the creative process: where the ideas came from, whether the work expresses who they are, and the lived experience of building the piece Where do writers locate authenticity in AI co-writing?. In this view authenticity is about having made something, not about what the result looks like.
That creates an odd gap. If authenticity lives in the process, readers mostly can't see it. One study found that readers couldn't tell AI-assisted writing from solo writing and didn't seem bothered by AI use. It didn't test whether they would care about process if it were disclosed Do readers value writing authenticity they cannot detect?. Writers face a similar split. People who use AI text say they don't feel they own or wrote it, yet they don't publicly credit the AI either. They treat it like an invisible ghostwriter Do people feel they own AI-generated text they use?. So the felt sense of authorship is real, but it's private, and it can differ from what writers say in public.
The theory-oriented notes add more by describing what AI text lacks. Each absence names an inner state that human writers bring to the page. One is a built-in appeal for the reader's attention: human writing reaches toward someone, while AI posts don't, which may explain why readers find them 'aloof' Does AI writing lack the internal appeal to attention that humans use?. Another is having actually lived what you describe. AI hotel reviews give themselves away partly because they make claims about experiences that never happened Does AI-generated text lose core properties of human writing?. A third is meaning what you say with something at stake, which is Habermas's idea of sincerity Can LLMs raise validity claims in Habermas's sense?. A fourth is having a body and being vulnerable, which one note argues no amount of language use can supply Do LLMs gain true linguistic agency through integration?. Put together, these give a fuller list than the interview study alone: intending a reader, having experienced, meaning it, and having something to lose.
The surprising part is that the process may not be fully invisible after all. AI fiction can be identified with 93% accuracy from narrative choices alone, such as how characters act and how time is ordered, with no help from writing style Can AI stories be detected without analyzing writing style?. That suggests the decisions writers make while working leave traces in a story's structure, even when readers can't name what they're noticing. The contrast with the model also sharpens the point. One view holds that a language model has no authentic voice underneath its personas; it is 'role-play all the way down' Does a language model have an authentic voice underneath?. Its reports about itself mostly echo human training data rather than any inner state Can language models actually introspect about their own states?, though some argue for granting it modest belief-like states Can we defend modest mental attributions to large language models?.
A caveat: the corpus has only one empirical study that asks writers this question directly, and the summary doesn't list specific internal states such as struggle, voice or intention one by one. The fuller list above is partly inferred from the philosophy notes about what AI lacks. It isn't what writers themselves reported.
Sources 11 notes
Professional writers co-writing with AI emphasize internal experience and the act of construction as central to authenticity, beyond the resulting text. Interviews with 19 writers revealed they define authenticity through source, identity, and the lived experience of making.
Hwang et al. found that readers could not distinguish AI-assisted from solo-written work and showed positive attitudes toward AI use. However, the study did not test whether readers would value process authenticity if disclosure occurred or if they could perceive it.
Two studies (n=30, n=96) found users do not feel they own AI-generated text, yet they refrain from publicly crediting the AI—treating it like an invisible ghostwriter. This gap between felt and declared authorship held even when AI text was personalized.
Human writing contains an appeal to the reader's attention as a fundamental property of communication itself. AI-generated posts inherit platform visibility but do not perform this internal appeal, producing the reported aloofness readers perceive — a structural absence, not a stylistic defect.
Research shows artificial text disrupts dialogic symmetry, context continuity, embodied authorship, and political situatedness. These are not surface flaws but structural absences—AI hotel reviews show 80%+ detection accuracy due to inherent falsity about personal experience distinct from human deception.
Show all 11 sources
Under Habermas's framework, LLMs cannot raise truth, rightness, or sincerity claims with genuine stakes. Without validity claims, their output fails to qualify as speech, making them non-speakers and non-interlocutors by definition.
Social grounding and linguistic agency are distinct properties. LLMs acquire more social grounding through integration into language communities, but remain categorically incapable of linguistic agency in the enactive sense, which requires embodiment and precariousness no amount of use can provide.
StoryScope achieved 93.2% accuracy separating AI from human fiction using only discourse-level features like character agency and chronological structure, retaining 97% of performance while eliminating stylistic cues. These structural choices resist humanization because they require rewrites, not surface edits.
Shanahan argues that base LLMs lack agency, beliefs, or preferences—the simulator is pure role-play with no underlying subject. Jailbreaking reveals the training data's full spectrum, not a hidden true self; even RLHF personas are performed characters, never realized quasi-psychologies.
LLM self-reports usually reflect human training distributions rather than actual internal processes. However, when a causal chain connects an internal state to accurate reporting—like inferring low temperature from output consistency—genuine lightweight introspection occurs without requiring consciousness.
Both robustness and etiological deflationist arguments beg the question against inflationism. A graded approach ascribing metaphysically undemanding states like beliefs and desires—while withholding consciousness claims—mirrors how we treat non-human animals.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- "It was 80% me, 20% AI": Seeking Authenticity in Co-Writing with Large Language Models
- Understanding Reader Perception Shifts upon Disclosure of AI Authorship
- Measuring and Mitigating Persona Distortions from AI Writing Assistance
- What Influences Readers' and Writers' Perceived Necessity of AI Disclosure?
- Penalizing Transparency? How AI Disclosure and Author Demographics Shape Human and AI Judgments About Writing
- The Assistant Erased You: Measuring Loss of Authorship Signals in AI-Mediated Communication
- The human-authorship halo: attribution bias in literary style evaluation by humans and AI
- The AI Ghostwriter Effect: When Users Do Not Perceive Ownership of AI-Generated Text But Self-Declare as Authors