When talking about binary units like bits and bytes, a kilobyte is always 1024 bytes, and a megabyte is always 2^20 bytes. When working with nonbinary items (e.g. kilogram, megaton), them they are decimal (10^3 10^6).
I have literally never heard anyone use these terms any other way. Nor have I ever seen anyone actually use the term kibibyte.
DRAM and SRAM capacities are in binary, but storage device capacities (from device manufacturers) are decimal. (I thought that storage device capacities were actually mixed, that is, decimal kilobytes, so a megabyte is 1000 * 1024, but I can't find a source for that.)
I think that Windows and Linux both use binary for disk space.
Most of the usage for kibibyte that I see is in articles advocating its use.
I agree, I haven't come across the usage of these terms either in my career, save for people who specifically wanted to make a point.
If it hadn't been in '99 already, I would say it's the language insanity which has plagued the world for the past decade.
While I understand the SI argument, computers "simply" work with the powers of 2, so a magnitude of 1000 "simply" never is the native unit here and the usual SI prefixes really shouldn't be an issue and never were. It only is a convenient excuse for harddrive manufacturers ;)
I have literally never heard anyone use these terms any other way. Nor have I ever seen anyone actually use the term kibibyte.