Raw LLM Responses

We could in principle make a neural network out of mechanical pieces that we could see. If we saw the machine turning and clunking when coming up with an answer and printing it out on a sheet of paper we would't call it sentient - we would just see it as a very intricate and strange mechanical process. If it doesn't have a metabolism, it isn't alive. The intense chatter about the neural network as a black box mystifies the whole thing. That's one direction we could go in - but then the essentially simple math that goes into neural nets should be mystified also. The other direction is a more grounded one where we use this new technology to see that our own intelligence isn't so mysterious after all.