Intelligence

People keep arguing about intelligence. What is intelligence? What is consciousness, and what does it mean to think?

I don't think there are many good answers, mainly because I prefer the EmptinessShunyata approach to understanding intelligence -- i.e., I don't think intelligence "really-really" exists as a phenomenon in nature, but is a piece of analytical overlay that we project onto the world. But there are still other views that I enjoy.

Emptiness / Shunyatic interpretation of Intelligence

This viewpoint is why I don't think intelligence is a good category.

To break down Intelligence according to EmptinessShunyata, intelligence is empty because it is made up of non-intelligent subcomponents, and is dependent on non-intelligent factors. Hence, no such thing as "intelligence" can ever be found in the universe; all you can find are the non-intelligent subcomponents of intelligence.

Going with the "Definitionalist approach to intelligence" below, intelligence must have an orientation towards a goal, and so it depends on orientation, which is a non-intelligent subcomponent which can be found even in a heap of rocks, which are oriented according to gravity. Also, another implication is that intelligence cannot exist without a goal, and so it depends on a goal for its own existence. In the absence of a goal, intelligence might seek for another goal (i.e. the "idle hands are a devil's workshop" problem), but that in itself is a goal; without a goal, we cannot say that intelligence exists.

Since these non-intelligent subcomponents can be mixed and matched in different ways, we can have things like Software which is somehow generally considered non-intelligent, but can still display some features of biological intelligence, and in many ways can even surpass intelligence (even without LLMs).

So, intelligence is not a very good category, and not a good descriptor of reality. A useful heuristic, sure.

Definitionalist approach to intelligence

A good definition of intelligence comes from William James, the father of American psychology:

Intelligence is a fixed goal with variable means of achieving it.
-William James

I like seeing intelligence this way because we can apply it to many different phenomena, while allowing us to transfer ideas between them.

For example, slime moulds can recreate the Tokyo rail system, because slime moulds follow similar patterns to humans when it comes to solving logistics & supply chain problems. This fits the definition of intelligence above. See Maze-solving by an amoeboid organism by Nakagaki et al

With LLM's, we are well past the human-centric view of intelligence. If we are to take the study of intelligence seriously, we should be able to apply it to all kinds of phenomena, not just humans. And this is where Michael Levin's chain of thought comes in:

I view intelligence as competency in navigating various spaces. These spaces can be... linguistic spaces and other things, it's competency navigating those spaces.
-MichaelLevin