ChatGPT’s Word Choices Now Echo in Human Speech

Researchers at an unspecified institution demonstrate that ChatGPT influences human language, finding words favored by the chatbot are appearing more frequently in spoken conversations. They analyzed 737,083 hours of podcast recordings from 824,634 episodes, focusing on unscripted speech.

The team observed a sudden increase in the use of words like “delve,” “showcase,” and “intricacies” immediately following ChatGPT’s release, linking this shift causally to the AI’s availability. A follow-up experiment with 496 participants confirmed that even brief interactions with ChatGPT led people to adopt the chatbot’s vocabulary, a change that persisted even after a distraction task.

This suggests humans internalize the lexical patterns presented by large language models. The findings raise questions about potential linguistic homogenization and the broad cultural impact of AI.