Sorry if it's bad form to take a discussion from a previous thread into the digest post, but I felt the need to address the point about mind-body duality. I am emphatically not a mind-body dualist, and I believe that most of the people who agreed with me weren't either. It's funny to hear you say that, because I interpreted your stance - that there is something called "theory of the world" that is endowed only to the human mind and can't possibly exist within a machine seemingly by definition - to be much more dualistic in nature. Obviously I acknowledge it isn't conventional dualism, but to me it has a similar taste.
Now that I am able to put a bit more time into my response, I want to give a bit more explanation of where I'm coming from. The first important thing is to separate the idea of consciousness - the subjective experience of being, the sensation of thinking - from the idea of intelligence and understanding - the ability to model and manipulate the world. Consciousness is obviously immeasurable, and completely undefined from a scientific perspective, so it is silly to talk about and I make no claims along that axis.
Intelligence, on the other hand, can be measured and demonstrated, and each generation of LLMs (usually less than a year apart) have shown dramatically increasing levels of it. Their mistakes are rapidly becoming rarer, and the situations in which they occur are becoming more complex. And don't forget that humans also make dumb mistakes all the time.
So then that leads me to my last question: What capability would these machines need to demonstrate before you admitted that they had something akin to a theory of the world? It's not a question of consciousness or experience or "mind" or other immeasurable things, it's a question about demonstrated ability to interact with the world.
I like when you publish too much. Keep 'em coming.
Sorry if it's bad form to take a discussion from a previous thread into the digest post, but I felt the need to address the point about mind-body duality. I am emphatically not a mind-body dualist, and I believe that most of the people who agreed with me weren't either. It's funny to hear you say that, because I interpreted your stance - that there is something called "theory of the world" that is endowed only to the human mind and can't possibly exist within a machine seemingly by definition - to be much more dualistic in nature. Obviously I acknowledge it isn't conventional dualism, but to me it has a similar taste.
Now that I am able to put a bit more time into my response, I want to give a bit more explanation of where I'm coming from. The first important thing is to separate the idea of consciousness - the subjective experience of being, the sensation of thinking - from the idea of intelligence and understanding - the ability to model and manipulate the world. Consciousness is obviously immeasurable, and completely undefined from a scientific perspective, so it is silly to talk about and I make no claims along that axis.
Intelligence, on the other hand, can be measured and demonstrated, and each generation of LLMs (usually less than a year apart) have shown dramatically increasing levels of it. Their mistakes are rapidly becoming rarer, and the situations in which they occur are becoming more complex. And don't forget that humans also make dumb mistakes all the time.
So then that leads me to my last question: What capability would these machines need to demonstrate before you admitted that they had something akin to a theory of the world? It's not a question of consciousness or experience or "mind" or other immeasurable things, it's a question about demonstrated ability to interact with the world.
I very much appreciated the quantity of posts this week.
Hoping the title of this post is referencing the Bright Eyes classic hit Something Vague