The focus of the phrase "jumping to conclusions" is on the process—the jumping—not on the particular conclusions that the person doing the jumping happens to land on. Thus, it is by no means odd to hear the caution against jumping to conclusions presented in roughly this form:
You may be right about X, but let's not jump to conclusions.
If the person is right about X, then the conclusion can't be deemed "bad"—but that still doesn't make the process good. The distinction here is reminiscent of the one that Socrates (in Plato's Meno) tries to make between false opinion (mistaken views arrived at by bad reasoning) true opinion (correct views arrived at by dumb luck despite bad reasoning) and knowledge (correct views arrived at by good reasoning).
If you jump to conclusions, you necessarily rely on dumb luck to find your way to true opinion rather than to false opinion—but you forgo any chance to arrive at knowledge. This is the essence of the warning, "Don't jump to conclusions."
John Locke, "Of the Conduct of the Understanding" (published posthumously in 1706, but written at least two years earlier), puts the problem this way:
In every question the nature and manner of the proof, it is capable of, should be considered, to make our enquiry, such as it should be. This would save a great deal of frequently employed pains, and lead us sooner to that discovery and possession of truth, we are capable of. The multiplying variety of arguments, especially frivolous ones, such as are all, that are merely verbal, is not only lost labour, but cumbers the memory to no purpose, and serves only to hinder it from seizing and holding of the truth, in all those cases which are capable of demonstration. In such way of proof the truth and certainty is seen, and the mind fully possesses it self of it ; when, in the other way of assent, it only hovers about it, is amused with uncertainties. In this superficial way, indeed, the mind is capable of more variety of plausible talk, but is not enlarged, as it should be, in its knowledge. It is to this same haste and impatience of the mind also, that a not due tracing of the arguments to their true foundation, is owing ; men see a little, presume a great deal, and so jump to the conclusion. This is a short way to fancy and conceit, and (if firmly embraced) to opiniatrey, but is certainly the farthest way about to knowledge. For he that will know, must by the connexion of the proofs see the truth, and the ground it stands on ; and therefore, if he has, for haste, skipt over what he should have examined, he must begin, and go over all again, or else he will never come to knowledge.
Incidentally, according to Christine Ammer, The Facts on File Dictionary of Clichés, second edition (2006), the wording has been around for approximately 315 years:
jump to conclusions, to To draw inferences too hastily from insufficient evidence. Also put in the singular (to jump to a conclusion), this cliché dates from about 1700.
which would put Locke's use of it near the front of the line.