Ce qui fait encore cruellement défaut à l'Intelligence Artificielle

Abstract

Artificial Intelligence still lacks an adequate theory of meaning. Maybe the obstacles we observe to the progress of AI are partially imaginary. We suppose for instance that there is an essential difference between opening a window « mechanically » because someone has asked us to do so and converting to a religion under the inspiration of a preacher. What if the phenomena were in fact of a similar nature: the power on a mind that a word has through its meaning ? The model of a gradient in memory associations generating discourse with emergent logical properties suggests we never intend to say the sentences we utter, but simply register - like anyone else - what are the words that our mouth utters. What we hear ourselves saying re-launches the affect dynamics underpinning our speaking, just as do sentences we hear when uttered by others.