Berlin: Artificial intelligence chatbots are primarily “trained” by being fed instantly with information from the Internet, partly from years of effort by some of the world’s leading people and thinkers.
But now people, including university lecturers and others called intellectuals, seem to be people who are unconsciously trained by AI.
A team of researchers based at the Max Planck Institute for Human Development Institute in Germany analyzed recent academic talks and podcast episodes and found what has been described as an increase in “measurable” and “sudden” uses of words “preferably generated” by ChatGpt.
The team argued that their work provided “the first large-scale empirical evidence that AI-driven language shifts are propagating into spontaneous speech communication beyond written text.”
After sifting through 360,000 YouTube broadcasts and double podcasts, researchers have discovered that since the launch of ChatGPT in 2022, the trend has been increasingly leaning in speakers to paint broadcasts with the words they regularly use in chatbots.
The team’s research suggests that the “linguistic impact” of AI extends beyond academia, science and technology.
According to the team, shifts are not only detectable in “scripted speeches” in lectures posted to YouTube, but can also be seen in more “conversational” or off-cuff podcasting.
A similar finding published in Science Advances shows that “a wide range of word analysis” in medical research papers published between 2010 and 2024 “a rapid increase in the frequency of words of certain styles” after AI tools became widely available.
Last year, a study led by the University of Tübingen in Germany found that “at least 13.5%” of biomedical papers had the characteristic of being “processed by LLMS.” – DPA
