Post #4123229

I'm going to make a prediction. LLMs will never be intelligent. They can't because they obviously lack some of the fundamental requirements. Don't ask me what these are(I could speculate), but the diagnosis is clear. All service tasks require intelligence. There's a lot of boilerplate, but underneath is intelligence, and the intelligence lies in the specific arrangement of the boilerplate. From this it follows there will never be large or even significant productivity gains from LLM use.