Temporal Flattening in LLM-Generated Text: Comparing Human and LLM Writing Trajectories

2026-04-13

AI Summary

This study investigates whether large language models (LLMs) can replicate the temporal evolution of semantic, lexical, and cognitive-affective structures observed in human writing. The authors construct a longitudinal dataset of human-authored texts spanning 2012–2024 and compare it against generation trajectories from three representative LLMs under both standard and historically conditioned settings. The work reveals, for the first time, a “temporal flattening” phenomenon in LLMs: their outputs lack the long-term semantic and affective drift characteristic of human writing. Using metrics based on drift and variance across semantic, lexical, and cognitive-affective dimensions—particularly under historical conditioning—the analysis shows that while LLMs exhibit greater lexical diversity, their semantic and emotional evolution is markedly weaker than that of humans. Notably, leveraging temporal variability alone enables distinguishing machine-generated from human texts with 94% accuracy and 98% ROC-AUC.

Technology Category

• Natural Language Processing: (Large) Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Large language models (LLMs) are increasingly used in daily applications, from content generation to code writing, where each interaction treats the model as stateless, generating responses independently without memory. Yet human writing is inherently longitudinal: authors' styles and cognitive states evolve across months and years. This raises a central question: can LLMs reproduce such temporal structure across extended time periods? We construct and publicly release a longitudinal dataset of 412 human authors and 6,086 documents spanning 2012--2024 across three domains (academic abstracts, blogs, news) and compare them to trajectories generated by three representative LLMs under standard and history-conditioned generation settings. Using drift and variance-based metrics over semantic, lexical, and cognitive-emotional representations, we find temporal flattening in LLM-generated text. LLMs produce greater lexical diversity but exhibit substantially reduced semantic and cognitive-emotional drift relative to humans. These differences are highly predictive: temporal variability patterns alone achieve 94% accuracy and 98% ROC-AUC in distinguishing human from LLM trajectories. Our results demonstrate that temporal flattening persists regardless of whether LLMs generate independently or with access to incremental history, revealing a fundamental property of current deployment paradigms. This gap has direct implications for applications requiring authentic temporal structure, such as synthetic training data and longitudinal text modeling.

Problem

• temporal flattening
• longitudinal text
• LLM-generated text
• writing trajectories
• temporal structure

Innovation

• temporal flattening
• longitudinal text analysis
• LLM vs human writing
• semantic drift
• cognitive-emotional trajectory