COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	@make(article)
C00005 ENDMK
C⊗;
@make(article)
@style(linewidth 6 inches)
@modify(hd2,use ux)
@style(indent 0,spacing 1.5,spread 0.5)
@style(topmargin 0.5 inches,bottommargin 0.2 inches, leftmargin 1.25 inches)
@style(indent 0)
@pageheading ()
@style(widowaction force)
@heading(ARTIFICIAL INTELLIGENCE, PRESENT AND FUTURE)

@Center(John McCarthy)

	At present it is possible to program computers to match
human performance in a number of restricted intellectual domains by
so-called expert systems.

	However, the thesis of this talk is that there are difficult
conceptual problems between the present state of artificial intelligence
research and the attainment of human level artificial intelligence or even
a reasonable degree of common sense.  I believe these problems will
eventually be solved, but whenever there are conceptual difficulties, we
cannot predict how long it will take.  If we knew precisely what these
problems are, perhaps we could solve them.  However, we know them by
their effects.  All present artificial intelligence systems are very
brittle, and it is easy to stump them by modifications of the problems
that would not give much difficulty to a human.

	In the talk I will present a number of mental qualities that
haven't yet been successfully programmed, some approaches that have
been tried and some that seem promising.