A Tale of Two Structures: Do LLMs Capture the Fractal Complexity of Language?
Ibrahim Alabdulmohsin, Andreas Steiner · 2025
Read A Tale of Two Structures: Do LLMs Capture the Fractal Complexity of Language? by Ibrahim Alabdulmohsin, Andreas Steiner (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-14924.txt.
Description
Ibrahim Alabdulmohsin and Andreas Steiner asked whether large language models copy the fractal shape of human language in a 2025 paper. Natural language shows self-similar complexity across scales, with links that stretch far across a text. The team measured such patterns in model output under varied heat settings and prompts.
Human text kept its fractal measures in a narrow band, while model text swung widely. That gap suggests fractal measures could flag a fair share of machine-written text. The findings held across three model families of different sizes. The authors also released a data set of over 240,000 model-written articles beside human matches. They frame fractals as a new lens for judging and detecting synthetic text.
Catalog record
| CREATOR | Ibrahim Alabdulmohsin, Andreas Steiner |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 161 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-14924