ADLASSNI@AWIIMC11.BITNET ("Adlassnig, Peter") (03/05/88)
Is it true that there are two main arguments against the feasibility of AI? 1) The philosophical and cognitive science argument (e.g., Dreyfus, Searle) 2) The computability and complexity theory argument (e.g., Lucas(?)) Could someone point out some relevant literature on the second point, please? Thank you in advance. Klaus-Peter Adlassnig Department of Medical Computer Science Garnisongasse 13 A - 1090 Vienna, Austria email: ADLASSNI at AWIIMC11.BITNET
gilbert@hci.hw.ac.UK (Gilbert Cockton) (03/30/88)
In article <8803051150.AA05897@ucbvax.Berkeley.EDU> ADLASSNI@AWIIMC11.BITNET ("Adlassnig, Peter") writes: > >Is it true that there are two main arguments against the feasibility >of AI? > .... Forget categories for the moment and come bottom up. Within formal semantics there are a whole set of problems which reduce confidence in the comprehensiveness of computational models of human beliefs and behaviour. Formal semantics is largely AI off-line, and has an intellectual and scholarly tradition which pre-dates the LISP bar of AI. I suggest you pick up the Cambridge University Press catalogue and chase up any Linguistics text with 'semantics' in the title. Most of these monographs and texts have consensus examples of problems for mathematical accounts of meaning, especially ones based on two-valued logics. Everyone in NLP should know about them. Basically, AI won't succeed until it cracks these problems, and there is no reason to believe that they will ever get anywhere near cracking them. The gap between mathematical accounts and reality remains too large. -- Gilbert Cockton, Scottish HCI Centre, Heriot-Watt University, Chambers St., Edinburgh, EH1 1HX. JANET: gilbert@uk.ac.hw.hci ARPA: gilbert%hci.hw.ac.uk@cs.ucl.ac.uk UUCP: ..{backbone}!mcvax!ukc!hci!gilbert