Understanding the argument -
A successful chess-playing computer would prove either that a machine can think or that chess does not involve thinking. - "A machine can think" means it'll apply "thinking" as humans, or it may not involve thinking like humans.
In either case the conception of human intelligence would surely change. - "conception of human intelligence would surely change," means say
1. If it can think - it means we may need to broaden our definition of human intelligence by including AI and not just limit it to biological beings (humans)
2. If no thinking is required - then we may need to evaluate whether we need to keep Chess as a game that needs thinking or not.
The conclusion is that the "conception of human intelligence would surely change." But what if there is a third possibility other than "thinking" or "no thinking." What if the computer uses probability or any other statistical method or anything but thinking (3rd possibility)? Then the conclusion that the "conception of human intelligence would surely change" will weaken?
The reasoning above is most vulnerable to criticism on the grounds that it does not consider the possibility that
(A) the conception of intelligence is inextricably linked to that of thought - out of scope.
(B) a truly successful chess program may never be invented - out of scope.
(C) computer programs have been successfully applied to games other than chess - "games other than chess" are out of scope
(D) a successful chess-playing computer would not model a human approach to chess playing - ok
(E) the inability to play chess has more to do with lack of opportunity than with lack of intelligence - out of scope