This is a bit like the question studio engineers ask themselves: i can spend all the money I have to give me a listening experience that is as flat and true as possible , but if 90% of people use apple airpods for listening to the result , wouldn’t i be better of prooflistening on that ?
And the answer mostly is that there is no single solution to use that pleases everyone , so you TARGET (not use ) what is regarded as the common default .
In monitors / tv / pictures , this is sRGB. So your OUTPUT should be set to sRGB .
Yes, someone might view your picture on a monitor that is terribly wrong in how it shows colors and brightness…
Or someone might view it on an awesome studio monitor, but has the wrong settings picked , or uses the wrong application, or …
You can’t account for everything all at once. Normal (sdr) pictures are believed to be sRGB if nothing else is known about them . And if an application doesn’t know what color management is, sRGB is assumed. And monitors are expected to try do show a good sRGB image out of the box (even better monitors often have a setting to use their full gamut and potential , to be sRGB accurate out of the box . Or they at least have a sRGB mode to LIMIT their potential but to be close to a safe default).
So , i would always target sRGB . If you know your audience (or the printer being used :P) , you can make more assumptions … and so maybe expect them to use color managed viewers , and then you can export in what you want as long as the profile is neatly saved in the file .
But for the web ? Or a phone ? Always assume sRGB .