From its very inception, the theory of evolution by natural selection has been misinterpreted by social reformers of all kinds. Opponents argue that it endorses a psychopathic 'nature red in tooth and claw' approach to social problems. Proponents argue the same (they just prefer to use Herbert Spencer's dictum of 'survival of the fittest').
The problem with these interpretations, of course, is that they are both instances of the naturalistic fallacy: taking a description of the world (an is) as if it were a prescription (an ought). David Hume showed, way before Darwin was even on the scene, that this is a logical problem: in an argument, a valid conclusion is supposed to be entailed by its premises, and you just can't reach an evaluative conclusion (one with an ought) based on purely empirical premises (ones with is's). Sorry, not going to happen.
So, taking a scientific view of the world cannot, without the aid of auxiliary hypotheses not themselves empirical, provide the basis upon which our moral judgments ought to rest (no matter what Sam Harris tells you). Evolution explains how things have worked, not whether that's how they should work (from a moral point of view).
Nevertheless, as the following lecture by philosopher Peter Singer shows, there are important and fascinating lessons that evolution can teach us about ourselves and our likelihood of reaching good or bad moral judgments, given our evolutionary makeup. And although these findings don't tell us what choices to make, they should make us aware of the biases we are likely to fall prey to without realizing it.
And if you want to learn more about the trolley problem, you could listen to the RadioLab guys talk with Joshua Greene on the subject.