"For the Good of All Mankind"

According to Kant we are supposed to always treat others as a mean and not an end. However, the utilitarian says that the greatest good for the greatest number is exercising good morals. In applying both of these philosophical approaches to human experimentation we run into a few problems.
Dr. Henry K. Beecher reviewed cases of violating human dignity and stepping out of bounds in human experimentation and came to the conclusion that the worst transgressions occurred when the experiment sought to benefit society and not the individual patient. According to Beecher, problems arise when humans are used as an ends to further knowledge instead of their own health. For the most part human experimentation has focused more on using people as an ends instead of a means, but should this be the case? The utilitarian view is difficult to accept because the benefits are difficult to measure. They usually occur in the future and putting an exact value to them is nearly impossible. But does this mean that we should limit the number of experiments conducted for the betterment of mankind. In a previous blog post I spoke about the Stanford Prison Experiment. It had obviously negative ramifications for the participants and the original intentions of the experimenter were ambiguous, but it has benefited society in some capacity. The participants consented under their own will and understood that there could be negative consequences. The benefits could be recognized when prisons restructure themselves so that inmates do not suffer the same malicious behavior. The number of people it could help could be infinite. But are we supposed to not allow these controlled experiments to happen?
I absolutely think that there is a line in experimentation, but if informed and willing people participate in controlled experiments, can't the knowledge gained be worth the negative affects? It isn't true across the board, but it can be true in many situations.