The study, which involved more than 10,000 participants, showed that using large language models (LLMs) and chat programs to obtain information can reduce the depth of knowledge acquired compared to learning through traditional internet searches. When participants learned a topic using AI-generated summaries, they provided advice that was more concise, less realistic, and more general than those who gathered information independently via Google searches.
Furthermore, readers found AI-assisted advice to be less helpful, indicating a decline in the quality of knowledge transfer, according to a report in the newspaper ” “. This finding was confirmed across various experimental controls, including scenarios where participants accessed identical facts but learned through AI summaries versus direct online interaction.
Experts also warn that over-reliance on AI for knowledge retrieval risks creating a knowledge void, where individuals offload memory and problem-solving tasks to machines, potentially weakening critical thinking and memory retention. While the ease and efficiency of learning with AI assistance are undeniable, this research emphasizes the need for a more intelligent and strategic use of AI.
Quick factual queries may benefit from AI tools, but developing deep, transferable knowledge requires active cognitive effort and engaged participation.
