If you sample at 1kHz, say, and square and then average the samples you get the mean-square - take the square root to get RMS.

If you just take the peak value then there are two problems:

1) you assume the waveform is sinusoidal - typically this won't be the case, since most loads are not linear (only heaters and tungsten lighting are linear really). Also the voltage may not be very sinusoidal (it should be but all the non-linear devices tend to affect it).

2) If the load fluctuates the peak value will consistently read high.

So it matters if accurate current measurement is your goal, but not if you just need a ballpark reading or want to detect changes.

To be accurate you still have to calibrate the TA12 and ideally measure the negative half-cycles too...