While I will agree that there is considerable "low hanging fruit" in terms of very legitimate science that can be done by sending robotic probes, there will reach a point in that research where having actual people physically there will make a whole lot of sense.
For the foreseeable future, the cost of sending humans beyond Earth orbit and bringing them back safely will always be too high compared to robotic probes. That the idea of disregarding the "and bringing them back safely" part is being discussed as a semi-serious proposal shows only that for some folks, manned space flight has nothing to do with rational cost/benefit analysis of scientific data gathering. For them it is instead some sort of near-religious holy quest. ("Some folks" being a general comment and not directed at you, Teancum.)
There is a reason why automated probes don't go running around Antarctica, even though sending people there happens at considerable expense.
Considerable expense?! Compared to even LEO, Antarctica is free. During the summer months there are 5,000 people there. Call me when we can support 5,000 people in LEO, much less out of range of resupply rockets. I should live so long. (Tell you want, I'd settle for living (in good health and the other usual long life caveats) until the population in orbit reaches 1,000, same as Antarctica in winter.)
With the distances involved, bandwidth for sending data can be a considerable problem.
The bandwidth problem does not change when a human is transmitting versus a robot. It is useless to station a human in space if they cannot report back, right? (Unless one assign some metaphysical value to "human eyes have seen such-and-such" -- but that's not a science mission, it's religious questing.)
Some local synthesis of the data...can take place in an automated fashion, eventually even that will eventually need to have somebody physically there to evaluate all of that data.
On what basis do you believe that a human could evaluate that data better than an expert system designed for the task? Especially one with continual (though time-lagged) feedback from Earth?
A good argument could be made that there is no need for human researchers to go for at least a century or more.
In a century, if we're still a technological society with a presence in space, we'll have robots capable of out-performing humans at any given task for which we might send humans into space.
Thrill-seekers (doin' it for the lulz), performance artists (doin' it for the "inspiration"), and religious questers (doin' it to martyr themselves doing what a robot could do more cheaply) aside, anything beyond Luna -- quite possibly, anything beyond LEO -- will always be for robots.