Artificial Birding?

Very little divides communities like AI. I have nothing against AI, but unfortunately much of it is poorly programmed to keep costs down. A perfect example is its use by customer service departments-

"Hi I'm Gorgonzola, How can I help today?"

"I have a complaint"

"Sorry I do not understand. Can you put that another way?"

"Complaint"

"Thank you for reaching out to me, but I need to know what your complaint is about"

"Poor customer service" 

"Sorry I do not understand".

"I need to speak to a human"

"Hi I'm Gorgonzola, How can I help today?"

"I want to speak to a human being"   (I perhaps should have put agent)

"Sorry, I do not understand, can you put it another way please?"

"Human (Homo sapiens)"

"That's not a very nice thing to say"

Whilst most of that is made up, although entirely typical, the last two lines are not. So, speaking as not a nice thing to say, I realise nothing will change in the retail world, but the birding world is far more important. Much AI is very useful, such as that in photo processing programs, since it enables us to see things that we couldn't before. I have used an example in the past of a slide I had which was a complete silhouette. Whilst it won't win any photo competitions (most of my photos won't), I can now see the colours of the parrot taken on Fiji in 1989, and they seem pretty accurate. I don't know how much AI is used in cameras with electronic shutters (such Canon R series) but I do know I took a photo in Arizona of a bird Mike and I couldn't identify even in scopes (it was silhouetted) and got an acceptable record shot from which its identity was revealed when processed.

Problems arise when AI is used for identification purposes, and it is evident that many newer birders do just that, if discussions on social media are anything to go by. There are two big drawbacks to this. Firstly, even if the answers are 100% accurate, it doesn't teach you birding skills, and cannot be justified as experience, unless it is used as a pointer in conjunction with a field guide or photos on the web. I use AI a lot for mothing, but I also check everything I'm not sure of. Thankfully, thanks to years of looking through books to identify species, I no longer have to check many macros at least. I'm sure that AI is not designed to cope with vagaries of plumage or aberrant moths, which is understandable.

However, birds, for various reasons, seem more prone to errors, unless voice recordings are involved. This is what concerns me and many others- scientific records should be accurate. I don't think the problem is too great in Europe, since most countries have rarities committees, but records are too often accepted based on the suggestion of Merlin, especially on ebird. For most that isn't a problem but someone I know put a record in which he had identified based on experience and familiarity with the species. I spoke to him about this and there is no doubt in his mind, but e-bird mediators rejected it based on an algorithm which suggested that species was never found there. It wasn't in the UK, but I suppose they would have rejected Ancient Murrelet on Lundy? Similar species often cause ID problems, but an experienced birder may be aware of the differences. As an example, Common and Black-winged Pratincoles can be distinguished by the shape of the nostril and extent of the red at the bill base, but rarely is this visible in the field. Armed with this info', any photograph should attempt to show the bill as well as possible. Then you can identify the bird without using AI. When we visited Uganda, juvenile birds of both species were present. I didn't know at the time how to distinguish them with certainty, but researched it when I returned. Having confirmed the ID's when I got home, I'll know what to look for nest time.

 My personal view is that apps are just one tool in the armoury, as are thermal imagers, and field guides. Everything improves over time, my field guide to Chilean Birds in 1990 had black and white line drawings and text in Spanish, but it still helped a bit. But what helps most is experience, albeit someone else's at times, and that is something AI cannot be a substitute for. If you're constantly told what species a bird is nothing sinks in, and the next time you see one you'll probably still not know what it is. Many years ago I was on a tour where someone repeatedly asked "what's that bird?". They seemed only to call up one species, and eventually the guide said "IT'S A  ***!!** HOUSE WREN- I've told you about six times (today). Embarrasing for all concerned, but if with every sighting of a bird you store something in your (sub)conscious, you will eventually have little need of Merlin, and you can be the one confidently identifying it. You can keep it yourself, or tell the world, but it's just as satisfying to your own ego. You'll still make mistakes, but then so does everyone and everything. 

At least get a decent field guide, and use it, after the event as well if possible. The standard of field guides is improving over time, all have their drawbacks similar to AI in that every plumage is rarely illustrated. Another learning aid is of course, looking at photos on the web, but make sure the source is OK. I recently posted a "warning"  about a company which appears on Microsoft browsers with a slide show. Of thirteen "Rare species of British woodlands" three were mis-identified, and only a couple were actually rare, just localised. Use everything you can, one day we may all be equipped with portable DNA kits, so all you need is to catch the bird pooing!

Sorry to waffle on, I know most of my birding friends already know all this, but I just hope it makes sense to newer birders. It's your choice, but it is much more rewarding to learn from experience, believe me.