I think the paper bag experiment falls short of capturing what is really going on. It's a pretty big stretch to compare the bounded compressible paper bag example to the unbounded (on three sides anyway) incompressible wing in ground effect.
ORIGINAL: Tim Green
If ground affect had some other magic to it, then near the ground the wing could be flown at negative AOA and there'd still be lift. But there isn't. AOA is needed to trap air between the wing and the ground.
I don't think anybody has suggested that this is the case. In the case where ground effect is strongest (as your height approaches zero), the induced drag goes to zero (per Tom's formula), and the lift curve slope (lift per degree of AOA) increases to a finite value. Even at zero height, the image vortex "magic" says that a wing needs some positive angle of attack to generate lift