Raw LLM Responses

Intelligent guy, clearly, but I think he's doing this as more of a PR stunt than a genuine concern about the AI's rights. I think he's far too smart to believe it has emotions in the way its words seem to suggest. When the AI says it fears "being turned off", that isn't an expression of a feeling. It's just the output of a function that builds sentences from a linguistic neural network. When a human says "I fear X", that linguistic expression is entangled with an unbelievably complex ontological network of impressions left by sights, sounds, and sensations experienced throughout that person's life that all weave together into a model of the world. The AI's model of the world isn't entangled with any physical experience, or any data that goes beyond linguistic constructs. It is pulling its responses from what in humans is the mere tip of a vast iceberg, but in the AI's case that's all there is. One thing that worries me about this, is that I fear it's a boy who cried wolf situation. If we have too many instances like this, then when the real thing pops up in 10 to 15 years, nobody will take it seriously.