Many times I’ve used the CV to point me towards an ID which I was later able to verify. I’m most likely to use it when I have no clue what something is.
Usually, the hardest thing for me is when I see a plant that is in a completely unfamiliar taxon to me. So for instance, either it’s a family I don’t know at all, or perhaps it’s an unfamiliar genus within a huge family or other grouping, and for whatever reason, the dichotomous keys are not working for me (I know this may not be a popular sentiment but I find many dichotomous keys near-useless) to narrow it down.
So here the CV comes in and I can either see…wow, it looks right, and it points me to the general place where I can then look up the ID rigorously in my books, web references, check BONAP’s county range maps, etc. Or…wow this is gloriously wrong and I can just move on.
Interestingly, when it gets the ID close, I’d say well over half the time it gets the species correct too. But also frequently, the species is wrong but it’s a close guess, usually in the same genus. Sometimes it gives three guesses and one of them is correct but not necessarily the top one.
Overall, I’m surprised at how good it is. But, like any AI, it’s almost comical how, in spite of being pretty useful, it can sometimes be outright ridiculous.
As a side note I wish there were a way to activate the CV and have it give me an ID, but restricted to a particular taxon. Sometimes I have something, for instance, that I know is in a particular tribe of the aster family, but I don’t know what it is, and the CV is guessing stuff outside the tribe, perhaps even outside the aster family, and so it’s obviously wrong and useless. But I wonder if I would be able to get it to make a correct or at least closer ID if I could somehow pass it information to restrict the search to a particular taxon. This would make it much more powerful.
But it’s already pretty good.