I see the iNaturalist’s artificial intelligence ID as serving 2 purposes. Each purpose needs a different language, and the first purpose needs a less specific level of taxon suggestion, that has no species level ID’s, but more genus, family, or order level suggestions.
One purpose of the suggested ID is to offer a “starting point ID” to an observer, within a group the observer doesn’t know well. They can then use this ID to allow those who know that group, or who know their part of that group, that the suggested ID is in, to find the observation in their searchs of those they want to identify. These more knowledgeable identifiers can then offer their ID’s, and potentially other information, such as answering the “tell us why” question, that is in the box below their ID. Especially if the first ID is not at species (or subspecies) level, it then doesn’t matter so much if any first ID is correct, but mostly that the observation was found by those who know the group best.
An AI suggested ID at a less specific level than species would be better, in that only offering that less specific ID is less likely to take a 3 to 1 vote to override. We could also ask how many specialists are there that only look at observations within their narrow area of specialty. Botanists that specialize in some subgroup of plants, are still likely to search through observations in broader groups of plants, than just those within their specialty, to be sure to find all of the observations within their area of specialty. A less specific taxon level ID, such as genus, family, or order, may be good enough for these identifiers to find it.
For iNaturalist to provide this starting point, so people knowing the group can find the observation, the language wouldn’t say “we’re pretty sure this is …”, instead it might say “this genus, family, or order seems like a safe place to start the ID process, so those who know the group best can find it, and offer their help getting it to the most specific, and accurate, ID they can. (and maybe, but if you think you have a good guess what species it is don’t be shy to suggest it. Others who know the group can confirm or correct your ID.)”
The second purpose I see of the iNaturalist suggested ID’s is to provide a potential list of taxa, preferably “seen locally”, that someone might research to then offer their own “suggested ID”.
Then the language for suggested taxa to research, to see if any of them seem to make a good fit, might be:
“Here are some taxa, seen nearby, that seem to look enough like the observation, to be worth checking out:”