The past 3 military aircraft I've flown (T-6, T-1, and KC-135) all have them. The T-6 also had a stick shaker, and the T-1 had a shaker plus giant red STALL lights and a horn when you exceeded something like 90% of the stall angle of attack. IIRC, the T-6 and T-1 had vane systems while the tanker uses a probe similar to the one shown by 35 AoA on the Hornet.
In my opinion, if you fly any sort of aerobatic light plane or large, transport-type aircraft, you should absolutely have one. We can argue over where to draw the line with respect to that. For aerobatic planes, I think it adds an extra dimension of safety, because you can literally stall an aircraft at any airspeed and orientation. Did I think it was possible for an aircraft to stall on the backside of a loop, pointed at the ground? Not until an instructor demo'ed it.
For transport aircraft, ESPECIALLY automated ones, it has multiple uses. I think the best way to crosscheck what the autopilot is doing is to reference the AOA. Like others have mentioned, I can tell at an instant if I am flying best range or endurance based upon what the gauge is telling me. I always cross-check to see if my approach speed corresponds with the appropriate AOA as well. This may create quite a few arguments, but in a situation like both the Air France and Colgan crashes, an AOA gauge would have saved the day. In fact, when I am practicing engine failures, takeoff-continued scenarios nearing the edge of the jet's performance envelope (heavyweight, high temps, etc.) in the sim, my crosscheck is mainly back and forth between the airspeed indicator and the AOA in order to prevent a stall at what is the normal takeoff attitude and get away from the ground.
Should it be in all planes? No, but I think you can make a very good case why it should be in a lot more.