There Are Monsters in Your LLM. By Machine Learning Street Talk
Large language models, like chatbots, can create confusion about their actual capabilities. People might mistakenly think these systems are more conscious than they are or fail to recognize their true abilities. Understanding these models requires careful conversation to avoid anthropomorphism and misinterpretation.
Highlights
embodied interaction with the world, uh, enables us to learn the kind of causal microstructure of the of the physical world. And the causal microstructure is is all about physical objects and the way they interact with each other and physical substances, liquids and gases and gravity and stuff like that