Bland new world: is AI making us all think the same?

PhD candidate Zhivar Sourati observes a homogeneity in recent academic research, claiming that artificial intelligence (AI) is to blame. Sourati and others are investigating how generative AI, which creates content like text and images, may be causing cultural and cognitive uniformity. This phenomenon resembles "McDonaldization," where fast-food industry efficiency and predictability have led to societal uniformity.

AI goes beyond merely disseminating information, actively shaping it. Studies suggest AI tools can flatten language, creativity, and cultural values, potentially influencing decisions, actions, and opinions.

The risk of AI homogenization is seen as an existential threat by computer scientist Emily Wenger. She warns that we might become less adaptable and suffer from groupthink if everyone relies on AI for communication. Wenger's research found that LLMs generated slightly more original ideas than humans, but their responses were more similar to each other than human responses.

AI can also reduce creativity. A 2024 study found that people who asked an LLM for ideas wrote more novel and enjoyable short stories, but these stories became more similar to each other. Another 2025 study showed that brainstorming with ChatGPT produced more creative ideas, but also reduced the diversity of ideas.

Meta-analyses have supported these findings, showing that AI homogenization is strongest for idea generation, especially in complex or constrained tasks. Recent studies have also observed similar effects outside laboratory experiments, such as a 2024 study analyzing over 400,000 scientific articles finding increased similarity in content and linguistic style after ChatGPT's release.

Another study found that using AI to correct grammar in human-written texts led to a decrease in variation in writing style, erasing personality, moral values, and demographics.

AI homogenization can also lead to cultural differences being muted. A study found that after ChatGPT's release, AI autocomplete tools increased writing similarity between individuals from different cultures, leading to more generic descriptions of cultural rituals and symbols. Additionally, AI models can affect the ideas of those who view them.

A study before the 2024 US presidential election found that LLMs were biased towards the initial Democratic nominee, Joe Biden, and affected the attitudes of Trump supporters who discussed politics with them.