Well, the good news is that "Test Circuit 2" in the spec sheet shows pin 6 (non-inverting input) grounded and the inverting input capacitively coupled. This indicates that the bias of the input stage will not be disturbed by bringing the non-inverting input pin to ground with your gain pot.
They say that the input resistance of the IC is 100K, so anything from 5K to 25K should be fine for the input pot. Generally speaking, higher pot resistance is better to minimize loading of the signal source. On the other hand, you don't want the pot resistance to be excessively loaded by the input resistance of the amplifier. With a 25K pot, its source resistance to the amplifier is highest at 50% rotation (assuming a low-impedance signal source). This places the two halves of the pot in parallel to (virtual) ground impedance for a 6.25K source impedance from the pot. This will not be excessively loaded by the 100K input resistance of the amplifier.
How are you going to use this amplifier? If it is going to be used in "the field," that is, being carried around and not in a fixed location, having the pots in the middle of the input cables is asking for trouble due to wire fatigue, accidental contact with the input conductor with resulting hum, etc. I would recommend putting the circuit and the gain pots in a project box available from many electronics suppliers like Digi-Key, Mouser, or possibly even Radio Shack. Metal provides better shielding, but for the low gain of this amplifier, a plastic box may be fine and no worse than hanging out in the breeze.
While 1/2 watt is certainly not going to give you much undistorted volume on home HiFi (do they still use that term?) speakers or a PA system, it will be adequate for personal listening in a quiet environment and it is a good introductory project.
Have fun.
awright