An Enactivist account of Mind Reading in Natural Language Understanding

  • 2021-11-11 12:46:00
  • Peter Wallis, Bruce Edmonds
  • 0

Abstract

In this paper we apply our understanding of the radical enactivist agenda toa classic AI-hard problem. Natural Language Understanding is a sub-field of AIresearch that looked easy to the pioneers. Thus the Turing Test, in itsoriginal form, assumed that the computer could use language and the challengewas to fake human intelligence. It turned out that playing chess and formallogic were easy compared to the necessary language skills. The techniques ofgood old-fashioned AI (GOFAI) assume symbolic representation is the core ofreasoning and human communication consisted of transferring representationsfrom one mind to another. But by this model one finds that representationsappear in another's mind, without appearing in the intermediary language.People communicate by mind reading it seems. Systems with speech interfacessuch as Alexa and Siri are of course common but they are limited. Rather thanadding mind reading skills, we introduced a "cheat" that enabled our systems tofake it. The cheat is simple and only slightly interesting to computerscientists and not at all interesting to philosophers. However, reading aboutthe enactivist idea that we "directly perceive" the intentions of others, ourcheat took on a new light and in this paper look again at how natural languageunderstanding might actually work between humans.

 

Quick Read (beta)

loading the full paper ...