I know this sounds great to most people but it demonstrates a very superficial level of thinking… I mean for sure an LLM is capable of asking questions, and if you set it up with real time “sensory” input it could generate constant reaction to that input… much in the way you are constantly being stimulated to react to your environment… I am not really sure what the distinction is between a biological brain and a predictive model or algorithm… I would ask you what you think your own brain is doing on a fundamental level.
I would actually argue that it is the most important question.
Surely the most relevant test of any intelligence is whether or not itself starting. Any classical description of an artificial general intelligence would surely require the thing to actually do work on its own. If an intelligence is of greater than human intellect but it has to be prompted in order to do anything, then it’s always going to be limited by what a human can think to prompt for.
Fuck if I know, but seems to me that intelligence is more than just reacting to stimulus. The problem is we’ve broken the Turing test. We’ve made a computer that can sound sentient, but clearly isn’t.
I know this sounds great to most people but it demonstrates a very superficial level of thinking… I mean for sure an LLM is capable of asking questions, and if you set it up with real time “sensory” input it could generate constant reaction to that input… much in the way you are constantly being stimulated to react to your environment… I am not really sure what the distinction is between a biological brain and a predictive model or algorithm… I would ask you what you think your own brain is doing on a fundamental level.
I would actually argue that it is the most important question.
Surely the most relevant test of any intelligence is whether or not itself starting. Any classical description of an artificial general intelligence would surely require the thing to actually do work on its own. If an intelligence is of greater than human intellect but it has to be prompted in order to do anything, then it’s always going to be limited by what a human can think to prompt for.
Fuck if I know, but seems to me that intelligence is more than just reacting to stimulus. The problem is we’ve broken the Turing test. We’ve made a computer that can sound sentient, but clearly isn’t.