The Words That Give Away Generative AI Text

image via WIRED

July 7, 2024, 10:00 AM

• •A new study estimates that at least 10 percent of scientific papers published in 2024 were written with the help of large language models (LLMs).
• •The researchers found that LLMs led to an abrupt increase in the frequency of certain style words, such as "delves,
• •showcasing
• • and "underscores.
• •
• •The researchers also found that LLMs were used more frequently in papers authored by non-native English speakers.

A new study has found that at least 10 percent of scientific papers published in 2024 were written with the help of large language models (LLMs). The researchers say that they were able to identify LLM-generated text by looking for a number of "marker words" that are commonly used by these models.

Read original article