I see three possibilities for this scenario:
1- The amplifiers have a current limited pull-up and pull-down with similar current limitations. In this case, they are essentially OTA's (operational transconductance amplifiers), and the output will be defined by the differential current output. In this case, the output will be at a point somewhere in between each individual output (if all is equal, it will be at the mid-point)
2- The amplifiers have a difference between pull up capability and pull down capability (classic 2-stage miller amp, for example). Assuming the output stage has a pull-down current source and a pull-up miller compensated gain stage, the amplifier that is trying to pull up will win, and the result will essentially be a "smaller current" selector (opposite if the current mirror is the pull-up, and will be a "greater current" selector) There is likely to be a range where the two currents are similar where there this will act like #1.
3- The amplifiers have a strong pull up and pull down capability. In this case, the output stages will fight with high current. Part damage is a possibility in this case, and the output will be determined by the part that is not damaged.
Of course, all of this assumes that the amplifiers are the same. If they are different, you may get different results (strongest output will win)