A: Recording levels
Since 24-bit allows to record at lower gain-levels to begin with you don't have to:
- use compression or extra talent with high dynamic sources (vocals!) *before* going digital
- watch hard for clipping during recording or verify the recorded material afterwards (time and concentration saver!)
- drive the analog-gain stages closer to their distortion point by staying close to 0 dB
B: Virtual instruments/effects
If a virtual instrument/effect introduces digital artifacts like aliasing (bad filter design) and you have to add gain-staging (compression, level, whatever) afterwards it's better to have those artifact in the lowest possible range.
With a 16-bit signal-path the lowest level any digital artifact can be is -96 dB, with 24-bit it's -144 dB = 48 dB more room to hide artifacts in. And since we are in the digital realm these are real values and not something limited by analog reality (up to the final DA conversion).