More and more people are joining the daily use of chatbots that work through artificial intelligence, a fact that is already being observed by scientists and psychologists, who agree when drawing the first conclusions about speech, writing and thinking of people in general, as reflected in a new study from the University of Southern California (United States)
Specifically, this research demonstrates that AI chatbots are standardizing the way people speak, write and think, ensuring that if this homogenization continues unchecked, there is a risk of reduce collective wisdom and adaptive capacity of humanity.
Greater real-world diversity
Table of Contents
The computer scientists and psychologists, who have published their arguments in an opinion article published in the journal ‘Trends in Cognitive Sciences‘ by Cell Press and collected by Europa Press, affirm that AI developers should incorporate more real-world diversity in the training sets of large language models (LLM), not only to preserve human cognitive diversity, but also to improve the reasoning ability of chatbots.
“People differ in the way they write, reason and see the world”
“People differ in the way they write, reason, and view the world,” contextualizes first author and computer scientist Zhivar Sourati of the University of Southern California. “When these differences are mediated by the same LLMyour linguistic style, perspective and strategies distinctive reasoning processes become homogenized, producing standardized expressions and thoughts for all users.
Confirmed: starting in 2026, the European Union will donate 413 million euros to Spain to produce fruits and vegetables
Within groups and societies, cognitive diversity drives creativity and problem solving, researchers say. However, cognitive diversity is declining around the world, as billions of people use the same chatbots of AI for an increasing number of tasks. When people use chatbots to polish your writing, e.g. it loses its stylistic individuality and they feel less responsible for their creative production.
“The concern is not just that LLMs shape the way people write or speak, but that they subtly redefine what counts as credible speech, correct perspective, or even good reasoning“warns Sourati.
Examples from the study
The team points to multiple studies showing that the results of LLM studies are less varied than human-generated writings and that they tend to reflect the language, values and reasoning styles of societies. Western, educated, industrialized, rich and democratic.
“Since LLMs are trained to capture and reproduce statistical regularities in their training data, they often sworkers represent the dominant languages and ideologies, Their results often reflect a narrow and biased slice of the human experience,” says Sourati.
Although studies show that individuals often generate more ideas with more detail when using LLM, groups of people They produce fewer and less creative ideas when they use LLM than when they simply combine their collective powers, the researchers note.
“Although people are not direct users of LLM, These will affect them indirectly.” insists Sourati. “If a lot of people around me think and talk a certain way, and I do things differently, I would feel pressure to align with them, because it would seem like a more credible or socially acceptable way to express my ideas.”
People’s opinions
Beyond language, studies have shown that, after interacting with biased LLMs, people’s opinions are more similar to the LLM that they used. LLMs also favor linear modes of reasoning, such as chain reasoning, which requires models to show reasoning step by step. This emphasis reduces the use of intuitive or abstract reasoning styles, which are sometimes more efficient than linear reasoning, the researchers say. They also point out that LLMs can alter people’s expectations, which can subtly change the direction of a person’s work.
“Rather than actively directing the generation, users often they get carried away by the continuations suggested the model and select options that seem ‘good enough’ instead of creating their own, which gradually shifts the initiative from the user to the model,” explains Sourati.
The solution
Researchers say AI developers should intentionally incorporate the diversity of language, perspectives and reasoning in their models. They emphasize that this diversity should be based on the diversity that exists in humans globally, rather than introducing random variation.
“If LLMs had more diverse ways of approaching ideas and problems, would better support collective intelligence and resolution capabilities of problems in our societies,” adds Sourati. “We need to diversify AI models themselves and, at the same time, adjust our interaction with them, especially given their widespread use in various tasks and contexts, to protect the cognitive diversity and ideation potential of future generations.”
