Upsampling audio streams

Do people upsample? If so, why? I’m not arguing one way or the other, just want to understand the logic.

(I’m coming from experience with graphics, and there, upsampling a raster file is rarely a good strategy, but I don’t know how direct the analogy is with audio)

My personal experience tells me that there is absolutely no advantage for a listener in upsampling.
Only reason to do it for the producers, engineers and so on is the possibility DURING this processings better to work with details.
My subjective impression as a listener: Upsampling is worse.

I convert my to my 44.1/16 files to 192/24. It started because some of the cheaper DACs have a fixed anti-alias filter. Nothing wrong with that but the manufacturers do not always pick the correct filter so I decided to upsample my files so I can select the filter. It works (I use Sox) but what I learnt was:

1 Should have done it at 96/24. 192 is a waste.
2 You must drop signal by at least 3.5dB. Otherwise it can sound hash.
3 It is really debatable if I can tell the difference.

To be honest it’s now part of my process.