If you plan to do any sound editing, mixing, mastering, I'd definitely go for 96k. It will minimize any rounding errors. Nothing there is likely to be audible but I'd rather have the extra bits earlier in the signal chain because you can't make the sound any more accurate than what you start with.
For playback, it really doesn't matter. Since the sampling rate (48kHz or 96 kHz) is twice the highest frequency that can be encoded, 48k already gets you to 24000 Hz. Unless you are a bat, you won't hear the difference.
If you are feeding the output to an external DAC, that might be a justification for using a high bitrate. The top end filtering that may be in the DAC will cause fewer anomalies if you are feeding in a signal with a higher bitrate. So-called brick wall filters can cause some ringing at nearby frequencies. Again, I doubt it would be audible with even a 48k bitrate but it will be way above audible frequencies using 96k.
Jim