GIB's defense is often bad, but should it be THIS bad?
Here it has seen all the diamonds, all the hearts, all the spade honors, I don't see any reason for it to think substantially more likely that partner started with CKJxx of clubs rather than declarer holding Jxx/Jxxx? Is the sampling algorithm just totally flummoxed when human doesn't really have his call? Plus the factor that it should probably weight setting contract over extra undertricks, though I guess it's probably hard to program it to determine whether a contract is "normal" or not.