The idea that the steering mode could make any measurable difference is silly, and ignores physics.
First, there is no physical way that it can make a difference when using AP. The car must be steered. The car must do all the steering. It must stay in the lane. The energy to do this is the same because all the work to steer must come from the car. Is there even proof that AP does anything different depending on the steering mode?
Now, when steering yourself, you do perform some of the work. Let's look at how much work that can be:
Let's say that you are steering constantly. You're putting 100 newtons of force (~10kg / 20lbs) into the wheel at a 250cm radius. You're a madman, spinning the wheel at 100 RPM the whole time. You've been training for this all your life. You were born to spin a wheel.
That's 30 watts of power.
So if you do this for one hour, you've saved 30Wh. Meanwhile, at 60MPH, the car has used 15,000Wh.
You've saved 0.2% energy. But this assumes you're doing a slalom all the way down the road. And you're very, very sweaty at the end.
In real driving? Yeah, it's not even 0.002%.
FYI, the steering rack is so low power that it's a 12V device, not a HV device. It can run off the small 12V battery in the car.
(For fun, let's expand this. A WORLD CLASS bicyclist can put out 300W for an hour. So 300Wh. Again, the car uses 15kWh in that time. So if you could get a world class cyclist to pedal in the back of your car and put that energy into the battery, the car would gain 2% efficiency.)
(Second expansion: The max power draw of the electric rack is about 50A, which is 600W at 12V. So if the rack was drawing peak power all the time, it would change your efficiency by 4%. Meanwhile, you can drive a car without power steering at all on the highway with one finger, so guess what the average steering load is? It's zero. Power steering is about assist at low speeds.)