In this paper, the author explains why Dreyfus’ “in-principle” critique of artificial intelligence successfully dismisses Traditional AI research but cannot sufficiently address (what the author calls) Parallel Distributed Processing (PDP) AI. The problem is that Dreyfus’ recent holistic thoughts run counter to his successful argumentative stance against Traditional AI. The author begins with the debates in- and theories, problems, and latest formulations of- Traditional AI. To flesh-out Dreyfus’ problem with this approach to AI, the author turns to Heidegger’s notion of understanding, making it clear that symbolic representations cannot account for understanding a thing as ready-at-hand, which is a crucial part of Dasein. But don’t PDP networks avoid the pitfalls of Traditional AI? Doesn’t Dreyfus just beg the question when he holds that “human beings are much more holistic than neural nets”? Contra late Dreyfus, the author believes that there is hope in PDP’s ability to exhibit non-formalizable behavior.
Gracilla, Nicholas K.
"Two Approaches to Artificial Intelligence: An Analysis of Heideggerian and Dreyfusian Critiques,"
Episteme: Vol. 4, Article 4.
Available at: https://digitalcommons.denison.edu/episteme/vol4/iss1/4