Chomsky has this horribly panned book called "Cartesian Linguistics" which captures very little of what Descartes is known for.
Back to Descartes, away from Chomskian interpretations : the Modern period of European philosophy begins with an epistemology of self-awareness. One tries to anchor the concept of "an algebraic I, the speaker/thinker/doer" to which subsequent ideas are attached.
This concept is particularly important in the modelling of data in anthropomorphic "intelligent" machines. One might say that fuzzyheadedness in building "AI" begins with a poor understanding of human data structures in the first place.
I just posted something on Reddit asking about "examples of frameworks for describing grammar, which have no concept of abstract speakerless/listenerless propositions, and which are instead grounded in the concept of the speaker/listener". I hope to learn something from the good public on this.
Meanwhile, I will simply bear it in mind as I develop the system which I am working on.
https://en.wikipedia.org/wiki/Cartesianism
Aside : I do think it's more accurate to speak of "AI" as anthropomorphic intelligence, than as artificial intelligence. Because, we don't currently have useful non-anthropomorphic concepts of intelligence.
No comments :
Post a Comment