AI Chatbots Are Changing The Way We Speak

A new study shows ChatGPT is influencing speech, with people using more formal language in everyday conversations.

A new study shows that AI chatbots like ChatGPT are influencing the words people use in everyday conversations. Researchers found that terms favored by ChatGPT (such as delve, comprehend, boast, swift, and meticulous) have surged in spoken language since the AI’s release in late 2022.

The study, conducted by researchers at the Max Planck Institute for Human Development, analyzed over 740,000 hours of YouTube academic talks and podcasts.  Using advanced statistical methods, they detected a clear shift in word choices, after ChatGPT became widely available.

The researchers argue that terms like delve, comprehend, boast, swift, and meticulous have experienced a significant increase in spoken language since the AI’s release in late 2022.

“These findings suggest a scenario where machines, originally trained on human data and subsequently exhibiting their own cultural traits, can, in turn, measurably reshape human culture,” said Hiromu Yakura, one of the study’s lead authors.

The researchers argue that ChatGPT generates its unique writing approach, favouring formal and refined vocabulary instead of basic human expressions. For example, it uses ‘‘delve’’ far more often than humans typically do. Scientists detected a significant increase of formal terms since ChatGPT launched, especially in academic and professional settings like science, business, and education.

Interestingly, the effect wasn’t uniform across all areas. Indeed, the term ‘‘delve’’ became more prevalent in science and business podcasts, yet its use didn’t change much in discussions about sports and religion.

The researchers warn that this trend could create a “closed cultural feedback loop”. This means that AI, trained on human language, starts influencing how people talk, which then feeds back into AI training data. The findings raise concerns regarding linguistic diversity, while also showing potential risks of AI-driven manipulation.

“Language is known to influence human cognition,” the authors write, noting that AI’s linguistic patterns could lead to “deeper integration into human thinking processes.”

The research team stresses the importance of conducting additional investigations about AI cultural effects, alongside monitoring these changes for proper oversight.

As reported by Scientific American, James Evans, a sociology and data science professor at the University of Chicago, said that analyzing word distribution is the right method for studying LLMs at this stage. He noted that as models mature, these patterns will become harder to detect, and scientists may need to examine broader linguistic features.