In my last post about the ARC-AGI benchmark , I explored how we measure progress in artificial intelligence through tasks that mirror human reasoning. The idea of machines “thinking” like us is fascinating, but it also reveals something deeper about our assumptions. We tend to imagine intelligence as something that resembles the human mind. Our tests, metaphors, and goals often reflect human-style cognition more than general intelligence itself. But what if intelligence does not have to look, sound, or behave like us at all? The Human Lens From the beginning, humans have measured intelligence through human eyes. Ancient philosophers like Aristotle believed that rational thought separated us from animals. Descartes later claimed that animals lacked minds because they could not speak or reason with words. Even today, we often treat conversation as proof of understanding. This perspective, called anthropocentrism , continues to shape how we study both consciousness and AI. Tes...