New study compares the spread of generative AI to cognitive viruses

An international research team has released a theoretical study, published as a preprint on arXiv, suggesting that generative artificial intelligence models such as ChatGPT and Claude behave like 'cognitive viruses.' Researchers emphasize that this term is not biological but describes technologies that spread culturally, influencing how humans think, learn, and manage information.

The study identifies utility as the primary transmission mechanism for these tools. As more individuals adopt AI for writing, information retrieval, problem-solving, and decision-making, it encourages widespread social adoption. This trend results in what the authors term 'cognitive offloading,' where humans shift part of their mental workload to external automated systems. As these technologies become deeply embedded in daily life, users may develop an increasing dependency on them, effectively altering human cognitive habits over time.

While the research frames this phenomenon through the lens of viral transmission, it focuses on the societal impact rather than a medical pathology. The work provides a conceptual framework for understanding how generative AI permeates society and permanently shifts human mental processes.