The limits of artificial general intelligence Artificial intelligence

can machine intelligent? can think ?

alan turing s polite convention
we need not decide if machine can think ; need decide if machine can act intelligently human being. approach philosophical problems associated artificial intelligence forms basis of turing test.


the dartmouth proposal
every aspect of learning or other feature of intelligence can precisely described machine can made simulate it. conjecture printed in proposal dartmouth conference of 1956, , represents position of working ai researchers.


newell , simon s physical symbol system hypothesis
physical symbol system has necessary , sufficient means of general intelligent action. newell , simon argue intelligence consists of formal operations on symbols. hubert dreyfus argued that, on contrary, human expertise depends on unconscious instinct rather conscious symbol manipulation , on having feel situation rather explicit symbolic knowledge. (see dreyfus critique of ai.)


gödelian arguments
gödel himself, john lucas (in 1961) , roger penrose (in more detailed argument 1989 onwards) made highly technical arguments human mathematicians can consistently see truth of own gödel statements , therefore have computational abilities beyond of mechanical turing machines. however, modern consensus in scientific , mathematical community these gödelian arguments fail.


the artificial brain argument
the brain can simulated machines , because brains intelligent, simulated brains must intelligent; machines can intelligent. hans moravec, ray kurzweil , others have argued technologically feasible copy brain directly hardware , software, , such simulation identical original.


the ai effect
machines intelligent, observers have failed recognize it. when deep blue beat garry kasparov in chess, machine acting intelligently. however, onlookers commonly discount behavior of artificial intelligence program arguing not real intelligence after all; real intelligence whatever intelligent behavior people can machines still cannot. known ai effect: ai whatever hasn t been done yet.




^ cite error: named reference turing test invoked never defined (see page).
^ cite error: named reference dartmouth proposal invoked never defined (see page).
^ cite error: named reference physical symbol system hypothesis invoked never defined (see page).
^ dreyfus criticized necessary condition of physical symbol system hypothesis, called psychological assumption : mind can viewed device operating on bits of information according formal rules. (dreyfus 1992, p. 156)
^ cite error: named reference dreyfus critique invoked never defined (see page).
^ cite error: named reference gödel himself invoked never defined (see page).
^ cite error: named reference mathematical objection invoked never defined (see page).
^ graham oppy (20 january 2015). gödel s incompleteness theorems . stanford encyclopedia of philosophy. retrieved 27 april 2016. these gödelian anti-mechanist arguments are, however, problematic, , there wide consensus fail. 
^ stuart j. russell; peter norvig (2010). 26.1.2: philosophical foundations/weak ai: can machines act intelligently?/the mathematical objection . artificial intelligence: modern approach (3rd ed.). upper saddle river, nj: prentice hall. isbn 0-13-604259-7. ...even if grant computers have limitations on can prove, there no evidence humans immune limitations. 
^ mark colyvan. introduction philosophy of mathematics. cambridge university press, 2012. 2.2.2, philosophical significance of gödel s incompleteness results : accepted wisdom (with concur) lucas-penrose arguments fail.
^ cite error: named reference brain simulation invoked never defined (see page).