Compare two worlds: in one, because of discrimination, there are fewer engineers, scientists, talented statesmen, teachers; in the other, without discrimination, there are more of these professions. The second world is more beneficial to everyone, even those of the advantaged groups of the first world. In the second world, people are more likely to discover beneficial technologies, economic approaches, medical breakthroughs, whatever.
In most universities there is a fixed number of PhD slots determined by the available budget. Then each person getting one of those jobs is blocking someone else from getting that job.
The economy expands to accommodate talent. Jobs are bounded by it. Magically double our collective intelligence, and what we understand as a "job" would cease to have meaning. Suddenly double the number of engineers, and our economy would boom.
But for the student finishing a master's degree, the only metric that matters is "PhD openings available this year". And short-term, job openings are zero sum.
Let's assume that all gender-unequal job distributions are due to discrimination (which is highly unlikely, btw), then by definition forcing equal outcomes (e.g. a 50% women quota) will harm the previously dominant group by reducing their share of the distribution.
The discussion shouldn't be if equality might harm some groups (it definitely does), but instead which level of harm we're willing to accept for the greater good (of more equality).