I guess it would largely depend on the FOV but I'd imagine that it wouldn't make that much of a difference for most applications that us amateurs are doing.
Most of the convection bubbles that an AO unit MIGHT be able to help with are several KM across but they're also far away. Would it help a FSQ with 4º FOV? Yes; but due to the under sampling it would only help with mount tracking errors.
Take a 10" F/4 Newtonian and ASI1600 setup, it has about 1º FOV and a guide star would be only a fraction wider than that. The AO unit is doing sub pixel movements over a relatively small and "slow" (0.5-2s) moving air mass that causes refraction, running a 3Hz you should be able to semi accurately model the movements. It is true that one side of the field will be different to the other as convection currents run both up and down depending on whether they're carrying warmer or cooler air but the law of averages say that it should show some improvement
I cannot quite remember the figures used in the original document I read but you'd only be looking at improvements of 1.8" to 1.7" or at a stretch 1.6" depending on what the atmosphere is doing at the time. Under generally poor seeing (strong jet stream) AO doesn't do much as it's a fast moving issue (need millisecond corrections at the wavefront).
My own theory is that it would work better for those of us at sea level where we have more atmosphere that we're imaging through and on nights of better seeing.