Except that is what they do. The scientists do some base trials so they know they're not giving you something that's stupidly toxic, then they sign up enough people who are similar enough in symptoms that the results can actually be analyzed, then they give them to you and hope you don't die. If the drug is awesomely good, they end the trial early and give it to more people.
If you just give people drugs randomly, cost doesn't go down. All that happens is that you stop being able to track how effective the drug candidates are, and have lots of people dead from drinking toxic snake oil.