Are you referring to the example in the link I originally provided? If so, it’s a really good question. Why would I decide to pick a species ID even though I am aware that it may be difficult (or that the current lore in the field assumes it is difficult) to distinguish this species from other, similar ones using photos?
I have partially answered this at the end of my long comment at the example. Sometimes, the observer may take into account additional information that makes one species the most probable choice. Things like seasonality, commonality, behavior, etc.
Other than that, choosing a tentative species ID can help with filtering observations in case you want to look at them side-by-side, which is an extremely useful tool for final identification. When you have a genus with 60 species but you know that a subset of your observations of this genus can only be one of two similar candidate species, then it helps to tentatively assign all of them to a single species. If you then filter out all of your observations of this species, you can see them side by side without having a lot of other species mixed in. It is then often much easier to figure out if you have only one species or perhaps two populations.