One way of making it clear how hard it is to ID each species that would be completely automated and harder to abuse would be to give each a “Percent of IDs disputed” or something similar. If every ID of a species is disputed (100%) that is a really hard thing to ID correctly. If IDs of that species are never disputed (0%) anyone might be able to ID it correctly.
This could be calculated as the ratio of number of observations ID as this that also have conflicting IDs over the total number of observations ever IDed as this. Potential identifiers could even search for things with initial IDs with less than x% disputed.