1. All of it. A difference in air pressure is just an abstraction for talking about forces on air particles. The description of what's going on with the screwdriver would be equivalent physically and much more intuitive if air pressure wasn't mentioned at all.
What's going on is the curved back pulls air around it, which deflects it downward, and that downward force on the air implies an upward force on the screwdriver. With the sharp edge back, this detaches the air flow instead, meaning there's much less deflection, and thus less force.
The air-deflection model is a much better intuition pump than the air pressure model. I think the historical dominance of the air pressure model is because it's easier to measure - just stick some air pressure gauges on surfaces, and you can calculate what that means for lift.
I guess it differs between people, what is easier to grasp :) . Pressure of air makes a real physical force (actually, pressure is defined as that total force divided by area). At the same time deflection downward, which implies an upward force, is correct but doesn't show the mechanism how actually that implied force is born. It is born out of the difference of pressure between top of the screwdriver and the bottom - at the bottom you have more or less undisturbed ambient pressure, at the top the flow reduces the pressure.
What was interesting to me is Ben's experiment with long enough cone which failed to get lift. Shouldn't be so, I think. The problem could be that his air jet is relatively narrow - so with a long cone, which makes an angle with the stream not the whole cone gets flown over, and not the whole area is under the reduced pressure. It would be very interesting to see if a wider air jet would still fail to lift the long tapered cone.
What's going on is the curved back pulls air around it, which deflects it downward, and that downward force on the air implies an upward force on the screwdriver. With the sharp edge back, this detaches the air flow instead, meaning there's much less deflection, and thus less force.
The air-deflection model is a much better intuition pump than the air pressure model. I think the historical dominance of the air pressure model is because it's easier to measure - just stick some air pressure gauges on surfaces, and you can calculate what that means for lift.