A new study published in Trends in Cognitive Sciences suggests that the widespread use of large language models (LLMs) such as ChatGPT, Claude, and Gemini is driving what scientists call “homogenisation” and “cognitive flattening.” Researchers warn that as people rely on AI to generate text, ideas, and reasoning, humans are beginning to sound — and perhaps even think — increasingly alike.
Zhivar Sourati, a computer scientist at the University of Southern California and one of the study’s authors, said the most obvious effects appear in writing. AI-assisted prose tends to favor certain words and stylistic patterns, such as overusing dashes or words like “quietly,” producing a recognizable AI-like tone. An analysis of Reddit posts, news articles, and academic papers found that a statistical measure of stylistic diversity fell by 20 percent after the release of ChatGPT in late 2022, indicating that people who have never met are converging toward similar vocabulary, tone, and sentence structure. “The stylistic individuality is flattened,” Sourati said.
The phenomenon extends beyond writing into collective creativity and problem-solving. Experiments with group brainstorming revealed that while individuals using AI tools generated more ideas than those working unaided, the overall diversity of ideas within the group decreased. In scientific research, LLMs appear to amplify individual productivity while narrowing the scope of collective exploration. Sourati explained, “AI-augmented work tends to move toward areas richest in existing data, automating established fields rather than opening new ones.”
Neuroscience research hints at even deeper cognitive implications. In one study, participants’ brain activity was monitored while writing essays. Those who used ChatGPT showed lower engagement with the task than participants writing unaided or using traditional search engines, and they remembered less of what they had produced. This suggests that reliance on AI may diminish the cognitive effort involved in processing, interpreting, and integrating information.
Despite these concerns, the authors note that not all standardization is negative. Clearer, more uniform language can improve communication efficiency. However, the potential cost is a reduction in diversity of thought. Studies in economics and psychology have long demonstrated that groups with diverse perspectives often outperform homogenous groups, even when individuals in the latter are highly capable.
Sourati drew parallels to earlier technologies, noting that concerns over memory loss from writing and knowledge offloading via the internet were not entirely unfounded. “With each of those earlier technologies, people still had to actively absorb information and apply it,” he said. “What is different with LLMs is that they generate the reasoning and the articulation for you.” The cognitive boundary between external tool and internal thought, he added, is becoming increasingly blurred.
With hundreds of millions of people interacting with the same handful of language models, the study raises urgent questions about the long-term effects on human creativity, originality, and problem-solving. The research underscores the paradox of AI: it can enhance productivity and clarity while simultaneously nudging humanity toward a more predictable, uniform way of thinking.
As AI becomes an ever-present part of daily life, researchers warn that preserving cognitive diversity and encouraging independent thought may become essential for sustaining innovation and the full breadth of human imagination.
