I’m noticing certain systemic problems with the species suggestions in the AI’s ID’s that are causing repeated mis-identifications.
An example I’ve been seeing a lot lately is recommendations of Sitka Spruce as “visually similar” to spruces in Eastern North America. Sitka Spruce only occurs in specific habitats in the west coast, and unlike blue spruce, Sitka spruce is not widely planted in landscaping. It doesn’t even survive well in most of the eastern US.
It’s getting annoying to correct many ID’s like this. Basically, this suggestion is creating a mess for other users to clean up, as inexperienced users select it because it “looks right”, never mind the fact that it doesn’t make sense based on range.
I also think we may be “mis-educating” people a bit. People who don’t know much about spruces see Sitka Spruce in the list and think that they need to check against it, and there isn’t anything to clue them in that it’s a west coast species that would be very unusual to see in the east, and that you don’t really need to focus on when starting out, if you live in the east.
I would like to be able to take some action to like, “fix” the algorithm to prevent this, for efficiency’s sake. But it seems like it would need to be done somewhat on a case by case basis.
I also would like the algorithm to distinguish between species that may be far out of range but are widely planted in landscaping (like Blue Spruce in the east) as these can escape cultivation and need to be checked against.
What is the best way to approach this?
This is just one example among many.