TechnologySaturday, October 24, 20262 min read

Why is AI generated language so hard to understand?

By Hacker News · Published by Everything Blog

In short

The research, submitted on October 21, 2026, focuses on a set of psycholinguistic features known to correlate with human processing difficulty, such as object relative clauses, long syntactic dependencies, and low-frequency n-grams. Using a novel corpus of AI-generated text across diverse domains, the study finds that AI models use these difficult features more often than paraphrases generated by round-trip translation. Specifically, the models exhibit a preference for inanimate subjects with animate verbs, low-fr…

Key points

The research, submitted on October 21, 2026, focuses on a set of psycholinguistic feature…: The research, submitted on October 21, 2026, focuses on a set of psycholinguistic features known to correlate with human processing difficulty, such as object relative clauses, long syntactic dependencies, and low-frequency n-grams.
Using a novel corpus of AI-generated text across diverse domains, the study finds that AI…: Using a novel corpus of AI-generated text across diverse domains, the study finds that AI models use these difficult features more often than paraphrases generated by round-trip translation.
Specifically, the models exhibit a preference for inanimate subjects with animate verbs,…: Specifically, the models exhibit a preference for inanimate subjects with animate verbs, low-frequency n-grams, longer dependencies, stacked nouns, and reduced object relatives in conversation.

Sources checked by our analysis pipeline: 1

Computer Science > Computation and Language

[Submitted on 21 Oct 2026]

Title:Why is AI-generated language so hard to understand?

View PDF HTML (experimental)Abstract:There is a widespread sense that AI-generated text has recently become harder to understand. In this study, we investigate a set of psycholinguistic features that are known to correlate with human processing difficulty (e.g., object relative clauses, long syntactic dependencies, low-frequency n-grams). Using a novel corpus of AI-generated text across diverse domains, we find that AI models use these difficult features more often than paraphrases generated by round-trip translation. Compared with human writing, all nine frontier and near-frontier models use more inanimate subjects with animate verbs and more low-frequency n-grams, and most also use longer dependencies, more stacked nouns, and more reduced object relatives in conversation; we also demonstrate substantial variation across model families. We propose and test a hypothesis as to why: while human language is optimized for successful communication, AI-generated text is increasingly optimized for other goals like math, code, and reasoning, in which there is a pressure for \textit{information density} rather than successful communication. We provide a proof-of-concept for this hypothesis by fine-tuning a language model under a length penalty, which leads to a divergence from human-like linguistic behavior. Our findings highlight the axes of misalignment between linguistic processing demands in humans and LLMs, and point toward possible remedies.

Additional Features

References & Citations

Loading...

Bibliographic and Citation Tools

Bibliographic Explorer (What is the Explorer?)

Connected Papers (What is Connected Papers?)

Litmaps (What is Litmaps?)

scite Smart Citations (What are Smart Citations?)

Code, Data and Media Associated with this Article

alphaXiv (What is alphaXiv?)

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub (What is DagsHub?)

Gotit.pub (What is GotitPub?)

Hugging Face (What is Huggingface?)

ScienceCast (What is ScienceCast?)

Demos

Recommenders and Search Tools

Influence Flower (What are Influence Flowers?)

CORE Recommender (What is CORE?)

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Original source: arxiv.org

AI Analysis

Tone

Neutral

In short

The research, submitted on October 21, 2026, focuses on a set of psycholinguistic features known to correlate with human processing difficulty, such as object relative clauses, long syntactic dependencies, and low-frequency n-grams. Using a novel corpus of AI-generated text across diverse domains, the study finds that AI models use these difficult features more often than paraphrases generated by round-trip translation. Specifically, the models exhibit a preference for inanimate subjects with animate verbs, low-fr…

Key Points

The research, submitted on October 21, 2026, focuses on a set of psycholinguistic feature…: The research, submitted on October 21, 2026, focuses on a set of psycholinguistic features known to correlate with human processing difficulty, such as object relative clauses, long syntactic dependencies, and low-frequency n-grams.
Using a novel corpus of AI-generated text across diverse domains, the study finds that AI…: Using a novel corpus of AI-generated text across diverse domains, the study finds that AI models use these difficult features more often than paraphrases generated by round-trip translation.
Specifically, the models exhibit a preference for inanimate subjects with animate verbs,…: Specifically, the models exhibit a preference for inanimate subjects with animate verbs, low-frequency n-grams, longer dependencies, stacked nouns, and reduced object relatives in conversation.

Topics

Technology

Was this summary useful?

AI Analysis

Tone

Neutral

In short

The research, submitted on October 21, 2026, focuses on a set of psycholinguistic features known to correlate with human processing difficulty, such as object relative clauses, long syntactic dependencies, and low-frequency n-grams. Using a novel corpus of AI-generated text across diverse domains, the study finds that AI models use these difficult features more often than paraphrases generated by round-trip translation. Specifically, the models exhibit a preference for inanimate subjects with animate verbs, low-fr…

Key Points

The research, submitted on October 21, 2026, focuses on a set of psycholinguistic feature…: The research, submitted on October 21, 2026, focuses on a set of psycholinguistic features known to correlate with human processing difficulty, such as object relative clauses, long syntactic dependencies, and low-frequency n-grams.
Using a novel corpus of AI-generated text across diverse domains, the study finds that AI…: Using a novel corpus of AI-generated text across diverse domains, the study finds that AI models use these difficult features more often than paraphrases generated by round-trip translation.
Specifically, the models exhibit a preference for inanimate subjects with animate verbs,…: Specifically, the models exhibit a preference for inanimate subjects with animate verbs, low-frequency n-grams, longer dependencies, stacked nouns, and reduced object relatives in conversation.

Topics

Technology

Was this summary useful?

Will AI expose the rickety infrastructure of the internet?
A recent vulnerability in PHP, a widely used programming language for building the internet, has raised concerns about the security of large language models and AI agents, emphasizing the need for immediate patching and increased vigilance to prevent widespread exploitation. This event underscores the importance of robust infrastructure and proactive measures to protect against AI-driven attacks.
6 min read
No Signs of AI Payoff Outside the Tech Sector
Despite significant investment and media attention, artificial intelligence (AI) has yet to yield tangible financial benefits for corporate America beyond the technology sector. Revenue per worker is increasing for the majority of companies (Mag 7), while it remains stagnant for the broader market (S&P 493). Profit margins are also rising for the majority (Mag 7) but are flat for the broader market (S&P 493).
1 min read
Source: Replit recently moved most AI spend from Anthropic to OpenAI, which gave 30%+ off in a deal worth $100M+/year, after Anthropic discontinued its discount (The Information)
### Briefing Replit, a software development platform, recently moved most of its artificial intelligence (AI) spending from Anthropic to OpenAI. This move resulted in a significant discount of 30% or more, with the total savings amounting to $100 million or more annually.
1 min read