Let me explain by analogy:
Say you've got a ship. If you perturb the ship by tilting it slightly starboard or port, and then remove the perturbation, it will return to sitting upright. Most ships are designed to have a stable equilibrium along the axis of the keel. Simply removing any perturbation will return the ship to the upright equilibrium.
Now, say you perturb the ship by applying a deflection to the rudder, tending to turn the ship to a new heading. When you remove the perturbation, the ship will straighten itself out, and stop turning, but will not return to its original heading. While most ships are designed to self-equillibriate to traveling straight, each possible heading is itself a stable equilibrium.
The "equality is treating people as equals" viewpoint assumes that the world behaves like the first case. You remove the discriminatory force, and you can assume that society will return itself to an upright equilibrium. But that's just an assumption. The world could very well behave like the second case. Even if you remove the discriminatory force, you can still be headed in the wrong direction!
The experience of at least two professions suggests that the world behaves more like the second case. In the 1950's and 1960's, there was extreme discrimination against women in medicine and law. In the 1970's through 1990's, both medical schools and law schools took affirmative measures to increase enrollment of women. And it was a smashing success: http://cdn.theatlantic.com/static/mt/assets/hua_hsu/assets_c....
The key takeaway, though, is that the affirmative action moved the society to a new stable equilibrium. Today, there is no real advantage to being a woman when applying to medical school or law school. But the enrollment is stable because the new situation is self-perpetuating.