I do a lot of recording and work with a lot of people who record live, acoustic instruments. I can't really speak for synthesizers, or sampled instruments. For the most part, most of the people I work with are in consensus on this area. Basically, going above 48k for uncompressed audio doesn't carry enough benefits to do it for every project. Generally, since people recording for these projects use 48k, the end project file also ends up being in 48k.
I have been asked to record above 48khz (the standard I deal with) twice. Once was at 96, once was at...192? That one was a pain because 192khz wav takes up so much space and transferring that monster took hours. The consensus with people I work with is that 48 is ideal for most purposes, 96 is great for blending remote sessions, and above that doesn't make enough of a difference to justify the increased resource requirements (hard drive space and processor power).
One minute of 48k wav is about 15mb. One minute of 192k wav is about 45mb (3 times the size). If I'm recording a multitracked brass section of 4 trumpets, 4 horns, 4 trombones, and a tuba (like I do for some clients), for a track that is 4 minutes long, I'm going to be playing about 2 to 2 and a half minutes for each track. This is 26 to 34 minutes of audio. At 48k wav, this is 390 to 510 mb of audio on the hard drive, and in RAM. For 192k, this is 1170 to 1530 mb of audio on the hard drive, and in RAM. That is much more resource heavy. The difference in sound quality isn't big enough to justify that.
If I were doing this for How to Save the World in 20 Minutes or Less (24 minute recording), you can do the math to find out how obnoxiously large that can get. I recorded that in 48k (and requested other performers to do the same) and the end result is definitely to my liking.
So that is my take on it. Although higher sample rates are appropriate for some instances, I don't ever really use them because 48-96 is appropriate for almost everything I'm involved in.