🔍 Read the full analysis: AI Text And Reader Fatigue: Six Patterns, With Practical Fixes on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
A preprint posted to arXiv on Oct. 21 compares writing from 14 language models with human text and identifies six linguistic patterns the authors say may make AI text harder to process. The researchers propose editing for clarity rather than relying on AI detectors, but they did not test whether readers actually understood the texts less well.
Three linguists have identified six patterns that appeared more often in writing from many advanced AI models than in comparable human text, in a preprint posted to arXiv on Oct. 21. The researchers say the patterns may make sentences harder to process, but they did not test whether readers actually understood the AI-written passages less well.
In the paper, “Why is AI-generated language so hard to understand?”, Thomas Hikaru Clark of New York University and Matias B. Martin and Kyle Mahowald of the University of Texas at Austin compared text from 14 language models with human writing. The material covered 120 casual chats, 45 coding sessions based on GitHub projects and 980 short persuasive essays. The human comparisons included subtitles, StackExchange posts, student essays and Wikipedia.
The six patterns were unusual word combinations, long distances between words that belong together, omission of the optional word “that,” strings of three or more nouns, nouns used as verbs, and descriptions of objects or problems as if they were alive. The authors reported that all nine frontier or near-frontier models in their sample used more unusual word combinations and more descriptions of inanimate things acting like living beings than the human comparison texts.
The source article offers editing suggestions for each pattern: use familiar wording, place verbs close to their subjects, retain “that” when it helps clarify a sentence, unpack noun strings, choose established verbs and avoid unnecessary personification. Its sample rewrites are the article author’s illustrations, not examples quoted from model outputs. One model sentence cited in the paper was from Claude Sonnet 5: “The guitar sitting in the corner because you’re still ‘deciding’ teaches you zero chords.”
The study at a glance
14 models, 3 settings, 6 patterns
Three linguists compared AI text with human text of a similar kind, then checked which hard-to-read patterns the models chose on their own.
The models
Frontier and near-frontier models are highlighted.
The comparison
- Chat vs. film and TV subtitles
- Code vs. StackExchange posts
- Essays vs. 2,934 high-scoring student essays (grades 6 to 12) and Wikipedia
- Control: every sentence translated into Japanese and back. If a pattern gets rarer, the model chose it over a plainer option.
- Cost: $251.19 in API fees for the whole data collection.
Held for all nine
More rare word pairs and more objects that act like living things than human writers.
Grew with training
In the Olmo family the effects rose from base to instruct to the reasoning model, Olmo Think.
Fixable
Claude Opus 5.5 scored lower than Opus 5 on 19 of 21 difficulty comparisons.
Source: Clark, Martin & Mahowald, Why is AI-generated language so hard to understand?, arXiv preprint, 21 October 2026 (not peer reviewed). Cost figure from the paper’s appendix A.6.
Powered by Thorsten Meyer AI
Editing May Beat AI Detection
The findings offer writers and editors a practical checklist for revising machine-generated prose without first determining who wrote it. If the patterns make a passage harder to follow, the suggested fixes focus on the text itself: make relationships between words explicit and use more familiar sentence structures.
The source article argues that editing is preferable to relying on AI detectors, which it describes as unreliable and capable of misidentifying human writing. That recommendation is not a result established by this study: the researchers compared language patterns and did not test detection tools or compare editing with detection. The paper’s findings are best treated as a set of patterns to examine, not a way to identify AI authorship or a guarantee that every flagged sentence is difficult to read.
Products Worth Considering

DJI Mini 5 Pro Fly More Combo with DJI RC 2, 1-Inch CMOS

Samsung 32 Odyssey G5 QHD Gaming Monitor

Bose QuietComfort Wireless Noise-Cancelling Headphones, Moonlight Grey
As an affiliate, we earn on qualifying purchases.
How Researchers Compared the Text
The researchers examined prompts and writing across different settings, then compared each category with human text of a similar kind. They also translated sentences into Japanese and back into English. According to the source article, this procedure was used as a control: if a pattern became less common after the wording changed while the meaning remained, the authors took that as an indication that a plainer alternative was available.
The sample included nine frontier or near-frontier models and five smaller open models from the Llama and Olmo families. The source article says the smaller models with lighter post-training wrote more like people, while the paper’s authors suggest that training for tasks such as math, coding and reasoning may reward dense answers over clear ones. That proposed explanation is an interpretation, not a cause demonstrated by the comparison. The paper’s appendix put data-collection API fees at $251.19.
“Why is AI-generated language so hard to understand?”
— Thomas Hikaru Clark, Matias B. Martin and Kyle Mahowald, as quoted in the paper’s title
Reader Comprehension Was Not Tested
The paper is a preprint and had not yet been peer reviewed, according to the source article. It examines English, and the researchers did not test whether actual readers found the identified passages slower to read or harder to understand. The claim that the patterns increase processing difficulty is linked to earlier studies, but this paper’s comparison does not itself establish their effect on comprehension.
The size and makeup of the model sample also limit how broadly the results can be applied. The study covers 14 models and particular writing tasks; its findings do not show that every AI system, prompt or generated passage uses the patterns. The authors’ explanation about training incentives remains a proposed account, rather than a confirmed cause.
Peer Review and Reader Tests
The next evidence to watch for is peer review and research that tests how people read passages containing these patterns. Studies with readers could help establish whether the features change comprehension or reading time, and whether the effect varies by writing task or audience. The source material does not give a timetable for peer review or further experiments.
For now, the practical advice is limited: writers can revise a sentence when its wording feels dense, but the six patterns should not be treated as proof that a passage was generated by AI. Whether the suggested edits measurably improve readers’ understanding remains an open question.
Key Questions
What are the six patterns identified in the paper?
They are unusual word combinations, long gaps between related words, omission of optional “that,” noun strings of three or more words, nouns used as verbs, and inanimate things described as if they were alive.
Does the study prove AI writing is harder to understand?
No. The researchers identified patterns and drew on earlier work about processing difficulty, but they did not test readers’ comprehension of the passages in this study.
Has the paper been peer reviewed?
The source article describes it as a preprint posted to arXiv on Oct. 21, 2026, and says it had not yet been peer reviewed.
Can the six patterns identify text written by AI?
No. The paper compares language patterns across sampled model and human writing; it does not establish a reliable authorship test. A pattern can appear in human writing too, so it is not proof that AI produced a passage.
What edits does the article recommend?
Use familiar wording, bring verbs closer to their subjects, keep “that” where it clarifies a sentence, unpack noun strings, choose established verbs and limit unnecessary personification.
Source: ThorstenMeyerAI.com
