IMO the difference between DLSS and FSR is over stated, i think when tech jurnoes have to blow the image up by 500% to count pixels they are clutching at straws.
They blow the images up to make it more obvious on a youtube compressed video which is fair enough. The differences are obvious in actual reality in things like temporal stability. MOst people will be watching these reviews on their phones, so being able to blow up with long zooms is necessary.
A prime example that anyone can test right now with is Starfield, an AMD sponsored game, yet enable FSR and you see unstable shimmering of background point sources, an issue that does not exist when using DLSS seconds later.
This is also an issue in Cyberpunk and plenty of others.
The bare fact of the matter is FSR is and seemingly always will be, 2 generations behind DLSS, it also uses software methods to do its job whilst DLSS uses hardware. Hardware upscaling will always be higher quality than software, regardless of how anyone tries to spin it.