To some extent it is all of those things. WAV and AIFF, for example, are formats used to encode uncompressed CD quality audio (Stereo, 44.100 Hz, 1440 kbps). Because they are as close as you can get to "pure audio" on your computer, they have the widest compatibility with different software and hardware, and so are important in developing professional audio, software, and in some professional archival situations.
AAC, MP3, MPC, Ogg Vorbis, and other lossy audio codecs all try to do the same thing: make transparent audio at ever-lower bitrates. There are some compatibility issues (Ogg Vorbis is open source, with one of the reasons behind its development being to create a patent-free codec, unlike MP3 and AAC), but each codec was developed primarily to try to out-do the others.
For any given person, the bitrate at which they achieve transparency will probably be different for each lossy codec, because the different codecs throw away different information. That's why there's really no "one-size fits all" standard in the lossy audio world, because each person's ears are different, and the different codecs produce different kinds of audio artifacts that different people may find more or less noticable. Buying better quality speakers can also expose artifacts that you wouldn't have noticed at lower bitrates.
Some codecs have been heavily tested, however. LAME's VBR presets, for example, have been developed and tested over a very long period of time and have been shown to be transparent for most people on most equipment. AAC has not been as extensively tested, but what testing has been done has shown it to reach transparency for most people somewhere around 192 kbps, depending on the music and equipment.
Powerbook G4 Mac OS X (10.4.3)