As a rule of thumb, servo up will make the steering more aggressive, and servo down will tend make it a bit numb. But...
That all depends on the suspension and chassis geometry. A BMI (DB10RR) has lots of rear grip, so running with the servo angled is essential to get a good balance. However, it doesn't rotate in the turn as well as...
...a CRC GenXL, which runs with the servo up. That car has very progressive steering, good rotation and good balance. But it isn't (in my view) as good as...
...an AE 12R5.1! This has the servo down, but its overall balance, steering and rotation is the best I've experienced in the current crop of cars.
So the answer is not as simple as saying one is different to the other in this way, and this is when you would use one or the other. It has to be placed in the context of the overall chassis performance.
The Ackermann effect is primarily governed by the position of the outboard steering ball stud in relation to the kingpin. Most cars use the AE front end, whose steering block geometry remains unchanged (L3 to R5) for many years. The CRC and Serpent geometry are slightly different.
The placement of the steering links fore and aft in the car will change the angles the wheels turn through, and have the effect of changing the relative angle of one wheel to the other. I'm never sure if that is the Ackermann angle (not according to the text book I have!), but this does have an effect on the steering response in the turn. However, so does the bump steer amount (shimming the ball stud up or not) when the servo is down!
My experience is that you set the servo in the same orientation as the top team guys, and then leave it there. If it was better to run a GenXL with the servo down, or an R5.1 with the servo up, then their team drivers would be doing it! HTH
