I'm a CS professor, so take everything I say as biased or whatever.
We all know and expect that most PhD students will not end up in academia. The best PhD students often aren't those who want to be academics, but ones who are just insanely curious. There aren't many times in life when you can work on something that just interests you in great depth and (hopefully) push forward what we collectively know. A PhD allows that. If you're doing it just for the qualification, you won't get nearly as much out of it, and may not finish at all.
So, what should you get out of a PhD? If the advisor does it right, the PhD student should learn what it is to do high quality scientific research. That is, ask a question, figure out a series of experiments that might allow that question to be answered, evaluate the data, and come up with conclusions supported by evidence.
It turns out this is insanely hard to teach, and I've never seen a new PhD student, no matter what their background, who can do this at the start. No undergraduate or MSc programme manages to teach this, though some students seem naturally somewhat better at it than others. We try only to admit those.
With a new PhD student, many of the ideas, most of the suggestions for experiments to run, and much of the interpretation of the results will usually come from the advisor.
As a PhD progresses, the student should be gaining experience in what it takes to frame a good experiment that might actually answer the question you want to answer. And they should be gaining experience on how to interpret results and what you can conclude from them. During this phase, much of my role is staring at graphs the student has produced, and asking "why does it do that?". An experienced student would already have an answer, a semi-experienced student will have noticed the anomaly, but not know how to figure out the cause, and an inexperience student will not have noticed anything is strange.
By the end of the PhD, a great student will be ahead of me in suggesting the experiments to run, and will have already run them before they come to see me. Only about 10% of PhD students reach this point.
In the end, people over-rate what the topic of a PhD is in the educational process. What most people get from a PhD is an apprenticeship in scientific method. Those are transferable skills, and the reason PhDs have endured as a qualification is we really don't know any other way to teach them.
Of course, no-one actually teaches professors to teach PhD students - we learned it by osmosis too - so the quality of professors in teaching these skills is pretty varied.
We all know and expect that most PhD students will not end up in academia. The best PhD students often aren't those who want to be academics, but ones who are just insanely curious. There aren't many times in life when you can work on something that just interests you in great depth and (hopefully) push forward what we collectively know. A PhD allows that. If you're doing it just for the qualification, you won't get nearly as much out of it, and may not finish at all.
So, what should you get out of a PhD? If the advisor does it right, the PhD student should learn what it is to do high quality scientific research. That is, ask a question, figure out a series of experiments that might allow that question to be answered, evaluate the data, and come up with conclusions supported by evidence.
It turns out this is insanely hard to teach, and I've never seen a new PhD student, no matter what their background, who can do this at the start. No undergraduate or MSc programme manages to teach this, though some students seem naturally somewhat better at it than others. We try only to admit those.
With a new PhD student, many of the ideas, most of the suggestions for experiments to run, and much of the interpretation of the results will usually come from the advisor.
As a PhD progresses, the student should be gaining experience in what it takes to frame a good experiment that might actually answer the question you want to answer. And they should be gaining experience on how to interpret results and what you can conclude from them. During this phase, much of my role is staring at graphs the student has produced, and asking "why does it do that?". An experienced student would already have an answer, a semi-experienced student will have noticed the anomaly, but not know how to figure out the cause, and an inexperience student will not have noticed anything is strange.
By the end of the PhD, a great student will be ahead of me in suggesting the experiments to run, and will have already run them before they come to see me. Only about 10% of PhD students reach this point.
In the end, people over-rate what the topic of a PhD is in the educational process. What most people get from a PhD is an apprenticeship in scientific method. Those are transferable skills, and the reason PhDs have endured as a qualification is we really don't know any other way to teach them.
Of course, no-one actually teaches professors to teach PhD students - we learned it by osmosis too - so the quality of professors in teaching these skills is pretty varied.