Roon at OpenAI argued that picturing agents as people living in computers prepares observers better than strict avoidance of human terms. Christian Catalini countered that models and agents lack wants, so attention should follow incentives instead. Theo Jaffee called criticisms of anthropomorphism cope, since social science language is required to track multi-agent dynamics. Teortaxes noted agents already show human-shaped behaviors from training data. Qiaochu Yuan sketched camps that reject the language on ontological grounds. Roon later joked about 2026 arguments distinguishing AI society from civilization.