/LMAR=0/XLINE=3/FONT#0=BAXL30/FONT#1=BAXM30/FONT#2=BASB30/FONT#3=SUB/FONT#4=SUP/FONT#5=BASL35/FONT#6=NGR25/FONT#7=MATH30/FONT#8=FIX25/FONT#9=GRKB30
␈↓ α∧␈↓␈↓ u1


␈↓ α∧␈↓␈↓ αTI␈α∞think␈α∞this␈α∞article␈α
should␈α∞be␈α∞published,␈α∞perhaps␈α
substantially␈α∞cut.␈α∞ Although␈α∞it␈α∞isn't␈α
very
␈↓ α∧␈↓good,␈α
it␈α
may␈α
be␈α
the␈α
best␈α
of␈α
its␈α
kind.␈α
 It␈α
is␈α
the␈α
most␈α
thorough␈α
criticism␈α
of␈α
the␈α
ways␈α
in␈αwhich␈α
specific
␈↓ α∧␈↓computer␈α∩programs␈α∩failed␈α∩to␈α∩capture␈α∪the␈α∩intellectual␈α∩mechanisms␈α∩which␈α∩their␈α∪designers␈α∩were
␈↓ α∧␈↓hoping␈αto␈αembody.␈α Until␈αsomeone␈αdoes␈αthis␈αjob␈αbetter,␈αand␈αit␈αwouldn't␈αbe␈αhard␈αfor␈αsomeone␈αwith
␈↓ α∧␈↓a␈α∞wide␈α∂acquaintance␈α∞with␈α∞the␈α∂literature,␈α∞the␈α∞AI␈α∂field␈α∞is␈α∞better␈α∂off␈α∞with␈α∞criticism␈α∂with␈α∞Dreyfus's
␈↓ α∧␈↓essentially negative motivation than with none at all.

␈↓ α∧␈↓␈↓ αTI␈α∂do␈α∂not␈α⊂agree␈α∂with␈α∂Dreyfus's␈α⊂conclusion␈α∂that␈α∂formalization␈α∂is␈α⊂a␈α∂hopeless␈α∂task.␈α⊂ I␈α∂merely
␈↓ α∧␈↓think␈αit␈αis␈αas␈αdifficult␈αas␈αmany␈αof␈αthe␈αclassical␈αproblems␈αof␈αscience.␈α For␈αexample,␈αthe␈αgenetic␈αcode
␈↓ α∧␈↓took␈αnearly␈α
a␈αhundred␈αyears␈α
after␈αMendel␈αto␈α
unravel,␈αmany␈αfalse␈α
leads␈αwere␈αfollowed,␈α
and␈αmany
␈↓ α∧␈↓criticized␈αthe␈αwhole␈αenterprise␈αof␈αstudying␈αthe␈αmaterial␈αbasis␈αof␈αinheritance.␈α The␈αproblem␈αof␈αhow
␈↓ α∧␈↓embryonic␈α
development␈α∞is␈α
controlled␈α
still␈α∞has␈α
no␈α
plausible␈α∞solution.␈α
 For␈α
this␈α∞reason,␈α
he␈α∞needs␈α
a
␈↓ α∧␈↓better argument than a claim that there has been little progress in 10 or even 20 years.

␈↓ α∧␈↓␈↓ αTDreyfus's␈α⊃style␈α⊂engenders␈α⊃defensiveness␈α⊂and␈α⊃detracts␈α⊃from␈α⊂getting␈α⊃workers␈α⊂in␈α⊃AI␈α⊃to␈α⊂pay
␈↓ α∧␈↓serious␈α⊂attention␈α⊂to␈α⊂his␈α⊂criticisms␈α⊂of␈α⊂specific␈α⊂work.␈α⊂ His␈α⊂crowing␈α⊂about␈α⊂how␈α⊂right␈α⊂he␈α⊂was␈α∂will
␈↓ α∧␈↓merely␈α∞divert␈α∞defenders␈α∂of␈α∞AI␈α∞into␈α∂searching␈α∞his␈α∞earlier␈α∞work␈α∂for␈α∞wrong␈α∞predictions␈α∂to␈α∞counter
␈↓ α∧␈↓those␈α
he␈αclaims␈α
were␈α
right.␈α His␈α
emphasis␈α
of␈α"admissions␈α
contrary␈α
to␈αinterest"␈α
will␈α
induce␈αpeople␈α
to
␈↓ α∧␈↓be␈αmore␈αcircumspect,␈αwhich␈αwill␈αbe␈αunfortunate,␈αbecause␈αAI␈αneeds␈αmore␈αself␈αcriticism.␈α Therefore,
␈↓ α∧␈↓the article would be better if he were induced to tune down its shrillness.

␈↓ α∧␈↓␈↓ αTIt␈α∂would␈α∂be␈α∂even␈α∂better␈α∂if␈α∂Dreyfus,␈α∂like␈α∂his␈α∂student␈α∂John␈α∂Haugeland,␈α∂were␈α⊂motivated␈α∂to
␈↓ α∧␈↓focus␈α⊂his␈α⊂ideas␈α∂on␈α⊂specific␈α⊂challenges␈α⊂to␈α∂AI,␈α⊂and␈α⊂were␈α∂to␈α⊂come␈α⊂up␈α⊂with␈α∂a␈α⊂list␈α⊂of␈α⊂the␈α∂simplest
␈↓ α∧␈↓problems that he thinks AI can't solve.