Artificial Intelligence: A Hermeneutic Defense

The field of Artificial Intelligence (AI), from its very beginnings in the 1950s, has been criticized for its name, as well as its ambition. Most of the debate concerns the possibility of artificial intelligence, and presumes there is indeed some thing which is intelligence; the only question has been whether or not artificial systems can be built which exhibit, or have, this thing. That is, the debate has remained for the most part within the rationalistic tradition. In this section, I would like to explore two alternative approaches to this issue. The first considers the consequences of viewing artificial intelligence as another metaphor for computing. That is, perhaps certain kinds of computing systems can be usefully viewed as being like intelligent beings in some significant way. An immediate consequence of this view, of course, would be a lowering of the aims and ambitions of the field. The claim that a system displays something like intelligence is surely much weaker than the claim that it is intelligent. However, I will reject viewing AI as a metaphor for another reason. The notion of metaphor presupposes that words do indeed have a core of certain meaning, their literal meaning. The metaphorical meanings deviate from the literal meaning in various ways, by removing some essential element of the definition of the term. By arguing that AI uses the term “intelligence” metaphorically, one would imply there is another, literal meaning. Instead, I propose taking the lessons of hermeneutics seriously, by accepting that terms have no static, context-independent, literal meaning. No one use of the word “intelligence” can make an exclusive claim to legitimacy. The focus of the debate about AI should not be its possibility, but rather the suitability of using the expression “artificial intelligence” in particular social or institutional situations for some particular purpose. Thus, whether or not it is helpful to speak