Does AI-generated text lose core properties of human writing?
Can artificial text preserve the fundamental structural features that make natural language meaningful—dialogic exchange, embedded context, authentic authorship, and worldly grounding? This asks whether AI disruption is fixable or inherent.
The "Hermeneutics of Artificial Text" paper identifies four properties of natural text that AI-generated text systematically disrupts or eliminates:
Dialogic symmetry — Natural text is created within a dialogical process: there is a mutual shaping between author and potential audience, between writing and reading. Artificial text eliminates this symmetry. The circumstances of creation — the process of shaping an argument, choosing means of expression — are completely different.
Communication structure — Natural communication is based on a dialogic scheme. Artificial text changes the communication situation: it is no longer symmetrical. One party (the AI) generates without the reciprocal dependencies that structure human communicative acts.
World representation — Natural text represents the world as the author experiences and understands it. Artificial text is created through processes that are technically, rhetorically, and cognitively different, which undermines the existing scheme of representing the world in language.
Context integrity — Natural texts always function in the context of other texts; they are embedded in a continuous social, political, and cultural context. This continuity is interrupted in artificial text. Context — including political and social context — is either changed or excluded.
These four disruptions are not surface-level deficiencies that better prompting can fix. They are structural consequences of how AI text is generated. The hermeneutic tradition treats text as a "condition of social processes" rather than a mere information container — which is why these disruptions matter beyond the aesthetic or stylistic.
Empirical evidence quantifies property 3 (world representation). When AI generates hotel reviews — writing as though it experienced a stay that never occurred — the text is linguistically distinct from both genuine and intentionally false human reviews: more analytic (higher function word rates), more emotional, more descriptive (higher adjective rates), and less readable. Classification accuracy between AI-generated and human reviews exceeds 80%. The AI text is "inherently false" about personal experiences — false by structural necessity, not by intent — and this falsity has a distinct linguistic signature compared to human deception, where falsity is intentional and thus linguistically strategized. See How does AI-generated false experience differ linguistically from human deception?.
What makes this sharp: the disruption is structural but the appearance is not. AI text looks like natural text, enters the same reading circuits, and gets interpreted using the same hermeneutic tools — which means the disruption propagates invisibly.
Source: Discourses
Related concepts in this collection
-
Does AI text affect readers the same way human text does?
If text is a condition of social processes rather than merely a container, does the origin of text matter to its effects? This explores whether AI-generated content enters the same interpretive and epistemic circuits as human writing.
the paradox: same circuits, different genesis
-
Why do ChatGPT essays lack evaluative depth despite grammatical strength?
ChatGPT writes grammatically coherent academic prose but uses fewer evaluative and evidential nouns than student writers. The question explores whether this rhetorical gap—favoring description over argument—reflects a fundamental limitation in how LLMs approach academic writing.
the metadiscursive evidence of what's missing in practice
-
Do LLMs develop the same kind of mind as humans?
Explores whether LLMs and humans share the intersubjective linguistic training that shapes cognition, and whether that shared training produces equivalent forms of agency and reflexivity.
the Habermas explanation of *why* these properties are absent
-
What makes linguistic agency impossible for language models?
From an enactive perspective, does linguistic agency require embodied participation and real stakes that LLMs fundamentally lack? This matters because it challenges whether LLMs can truly engage in language or only generate text.
the enactive cognitive science framing: embodiment, participation, and precariousness are constitutive of the four properties the hermeneutic view identifies as absent — two frameworks converging on the same structural gap from different disciplines
Click a node to walk · click center to open · click Open full network for a force-directed map
Original note title
artificial text eliminates four foundational properties of natural text