A tweet by @HowToPrompt__ states that Harvard published a paper titled “Large-Language Models as a Cognitive Virus.” It says researchers from Harvard and the Santa Fe Institute examined LLMs via evolutionary biology, complex systems, and epidemiology, and concluded that language models spread like viruses. Visible replies include Robin Hanson noting jagged LLM abilities and limits on deeper conceptual issues, Timnit Gebru disputing characterizations of prior work, Valerio Capraro referencing Geoffrey Hinton comments, and Aella stating concern over AI takeover risks.