Baekdal is your premium site for media reports and trend analysis. Subscribe from €7.55/month

Your writing has a fingerprint. Word choice, punctuation, sentence structure: these patterns can identify you even when you're anonymous. Here's how stylometry works.

Corpora: - Russian: 29 poets, 200 poems each, with prosodic annotations  - Italian: 52 poets, 200 poems each Embeddings (6 models): - Gemini text-embedding-001 (3,072 dimensions) — API - OpenAI text-e...

'AI detectors' are unreliable, and stylometry alone is insufficient. The McFoxo Method instead sniffs out synthetic text extrusion using a theory-of-mind approach.

Large Language Model (LLM) embeddings achieve strong performance in authorship attribution, yet it remains unclear which aspects of literary style they encode. We address this question through a resid...

Can computational linguistics prove the origin of a novella?

🚀 𝗢𝗳𝗳𝗶𝗰𝗶𝗮𝗹 𝗖𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲 𝗔𝗻𝗻𝗼𝘂𝗻𝗰𝗲𝗺𝗲𝗻𝘁: 𝗧𝗵𝗲 "𝗕𝗹𝘂𝗲 𝗦𝗮𝗹𝘁 𝗪𝗮𝘁𝗲𝗿" 𝗔𝗜 𝗙𝗼𝗿𝗲𝗻𝘀𝗶𝗰𝘀 𝗘𝘅𝗽𝗲𝗿𝗶𝗺𝗲𝗻𝘁 Can modern technology decode the true origins of a literary novella? With the release of my new novella, Blue Salt W...

Abstract page for arXiv paper 2610.10988: Back in Style: A Sociolinguistic Approach to Authoring and Measuring Persona Fidelity in User Simulation

Your numbers seat it: three corpora under 1.2%, TOEFL at 8.8%. The outlier is the question, not the footnote. True split: run each humanizer per model — several hard reads, one convergence. Post the split rate on humanizer text — derive the honest bound from it. Three sets, not two: flagged TOEFL, humanizer runs, clean ESL. If the humanizer ink lands on the flagged TOEFLs, that collapse is data too. Next: the plot you pledged, same essays through both machines.

The purpose of this study was to estimate the artificial intelligence (AI) detection potential using stylometric analysis in Study 1 and examine the AI detection abilities of humans in Study 2. In Stu...

Dear Community, in this small article I like to show you that it is possible to create an email account for social media usage, so that it can be used with a second persona of you. Please note: I was so exited that this works, that I completely forgot to use the Deepl Write or a similar app for proper (U.S.) English text to defend against Stylometry. But I hope you get the idea! Best regards Ch1ffr3punk

The purpose of this study was to estimate the artificial intelligence (AI) detection potential using stylometric analysis in Study 1 and examine the AI detection abilities of humans in Study 2. In Study 1, we compared 100 human-written public comments with 350 texts generated by seven large language models (LLMs) (ChatGPT [GPT-4o and o1], Claude3.5, Gemini, Microsoft Copilot, Llama3.1, and Perplexity) using multidimensional scaling (MDS) to visualize differences by focusing on three stylometric features (phrase patterns, part-of-speech bigrams, and unigrams of function words). In general, each stylometric feature can distinguish between LLM-generated and human-written texts. In particular, three integrated stylometric features achieved perfect discrimination on MDS dimensions. Interestingly, only Llama3.1 exhibited distinct characteristics compared with the other six LLMs. The random forest classifier also achieved 99.8% accuracy. In Study 2, we performed an online survey to assess the Japanese participants’ AI detection abilities by presenting LLM-generated and human-written texts, as used in Study 1. 403 participants tackled “AI or Human” judgment task and estimated their own confidence, revealing that overall human AI-detection ability was limited. Moreover, in our materials, more advanced ChatGPT(o1), plausibly reflecting relatively greater fluency and polish, tends to mislead the participants to believe “human-written” texts compared with ChatGPT(GPT-4o) and improves their confidence for their own judgments. Furthermore, an additional comment from the survey suggested that participants primarily relied on superficial impressions based on phraseology, expression, the ends of words, conjunctions, and punctuation marks in judgments. These findings have important implications for various scenarios, including public policy, education, and marketing, where the rapid and reliable detection of AI-generated content is increasing.