A study from the University of Copenhagen has warned that relying too much on AI chatbots for information can lead to a “knowledge collapse.” This happens because users are exposed to fewer different facts and viewpoints compared to using a regular web search like Google.
AI chatbots, such as ChatGPT, Gemini, or Claude, have become popular tools for answering questions. People use them for everything from cooking recipes and email advice to understanding complex topics like rising interest rates. These chatbots use large language models to understand questions and provide answers.
Researchers at Copenhagen’s Department of Computer Science tested 27 AI chatbots across 155 topics related to 12 countries, creating over 1.7 million answers and analyzing millions of claims. They found that every AI chatbot provided more similar responses than a simple Google search. Even the most diverse model tested, OpenAI’s GPT-5, gave at least 18.7% fewer varied information compared to Google.
The study highlights that AI chatbots are not just changing how we access knowledge but also limiting the range of knowledge available. Users often receive the same information repeatedly, reducing exposure to different perspectives. This could narrow our understanding of various topics over time.
Dustin Wright, a researcher at Aalborg University and lead author of the study, explains: “AI chatbots give users more similar information than Google across all topics we studied. People are exposed to the same information again and again. AI chatbots are altering not just how we find knowledge but also what knowledge is available to us.”