Returning - somewhat reluctantly - to the debate on the value/purpose/nuisance factor of AI here on iSpot.
I personally don’t find it that intrusive. But it is no longer amusing (my Blackbird identified as a Mallard, and that Goat as a Wolf were worth a bemused chuckle) either. And, given the poor accuracy (Plantnet - I haven’t noticed a Fastcat effort for a while), I tend simply to ignore the offerings.
What I do find annoying:
the way they inflated my Changes Tracker list
the lack of any rationale for their choices. (I have wittered on before about the fact that they are never going to give any, since this is effectively their stock-in-trade)
the consequent failure to provide any chance to improve the human user’s skills
the likely bolstering of the belief that, because it comes out of a computer, it must be right.
Perhaps a poll should be held to see if users want to continue seeing the results by default? The process (the basic purpose of which, I assume, is providing learning materials for the AI) could continue ‘behind the scenes’, and anyone who is finding it interesting or valuable could continue access the data via a menu option.
It would be nice to have the AI as a single line in the comments that you could click on if you wanted to see the results or simply ignore and it would not take up space. However there are more important issues to deal with such as updating the global/ZA species dictionary and various underlying parts of the code being kept up to date etc and it is these things that the programmer needs to concentrate on, these are huge tasks in themselves. Chris has been doing a whole load of other things behind the scenes recently which are required just to keep the site alive (and which hopefully the general users won’t have noticed at all) so the issues with AI are not a priority at present. Perhaps unfortunately nothing stands still in computing, if you are not constantly updating and changing (even if the users think nothing has changed) then things fall to bits and when you have a complex site this is a large ongoing effort.
It seems a simple move, to me, to simply pause PlantNet’s activity.
It has been running long enough for evaluation
The App is good but the performance in iSpot is not up to the iSpot Process (of getting an ID)
So, miked, my earlier question … quote=“JoC, post:19, topic:2015”]
Would reporting PlantNet Suggestions that are way off, taxonomically speaking, be appropriate? I mean, would you remove them?
[/quote]
Doing that i.e. removing suggestions would make it much more difficult for us to get an error rate (just because of the way things are set up in database etc).
We had always intended to feed back to the AI system authors the suggestions from our human users so that they could examine “failures” and possibly improve their system - but the project ended, as did the funding.
Given the update by Chris, that the project has ended, then perhaps Dejay’s suggestion that PlantNets activity be paused - or even ended - is worth considering.
Depends on the wording of the project contract - sometimes project partners are expected to continue “publishing” for a while after the project finished - often up to three years. And if it remains in place, there’s a greater chance of additional funding from a follow-up project. It would probably be better for all concerned to have a personal setting along the lines of “hide AI contributions” in everyone’s user account.
If so, one would want to hide them in the Changes Tracker as well as in comment trails. (I find them more of a nuisance in the Changes Tracker, but there is the point that they also ruin the point of the comments present marker on the observation.)
If so, one would want to hide them in the Changes Tracker as well as in comment trails.
Excellent point! One little change would have lots of consequences and in fact would make the queries that build the Changes Tracker that little bit more complicated - and slower.
plenty of active users would agree to that. I feel a temporary Pause for all the reasons apparent IN comments (about PlantNet specially) show
I still do not understand WHY the Trials are not running in the Test Site where a few of us could be nothing but constructive and where it would not be so intrusive. After all it is NOT the response of users that the Developers ‘want’ but where they
