How to tell if a book was written by AI, and how the use of AI is disguised
An article in The Conversation by Javier Arroyo Gallardo (University of Alcalá) explains how AI-generated text is detected and how it is concealed. AI writes token by token, choosing among words whose probabilities are tied to their frequency in training texts, leaving traces such as common vocabulary, filler words and didactic formulas. Stylometry, perplexity and detectors specifically trained for the task, such as Pangram, are the strategies described. The text mentions the case of Thélyson Orélien's novel, withdrawn from a literary competition over suspected AI use.
In 40 seconds
- AI generates text token by token, with probabilities tied to word frequency in its training texts
- Stylometry and perplexity look for uniform patterns that reveal AI-generated text
- Detectors trained for the task, such as Pangram, are the best existing strategy according to the article
IASummary generated by artificial intelligence from a story by BioBioChile. It may contain errors: the original source prevails.




Comments
Be respectful. Comments reported by several people are hidden until a moderator reviews them.
You do not need an account to read, only to comment. Sign in with a code sent to your email or with your Google account. We only show the name you choose, never your email.