With all the discussions about the use of AI on iSpot, it got me thinking about where wildlife watching and in particular bird watching is heading. How long before spotting scopes come complete with gps and AI. The bird ID AI that iSpot uses is nearly always correct - Fastcat. How long before your scope automatically counts and submits records of all your sightings? Personally, I think that would take the fun out of birding, and I wouldn’t use it. But from a scientific (or conservation?) point of view, it would be invaluable. (If Merlin were included, too, you’d have a whole heap more records although some of them would be spurious.)
If anyone has the inclination, I’d be interested to know where you feel technology will take wildlife watching/recording in the coming decade or two. (I’m trying to avoid a discussion about whether or not humankind will continue to progress into the longer-term future.)
I’m sure I read somewhere that AI is already being used in bird counting.
Think it can be used in conjunction with drones for seabird counts.
Can just imagine the fixed video cameras at certain birdy sites being linked to AI to get a constant stream of species and numbers that are present, and the huge numbers of extra visitors that would turn up if the AI makes a mistake and spots a very rare species by mistake.
The question “I’d be interested to know where you feel technology will take wildlife watching/recording in the coming decade or two.[in relation to AI]” is somewhat similar to a current call for funding so the government also wants to know the answer.
I used Merlin yesterday for a survey.
I left it running, on my phone, for half and hour in my quiet woodland. It registered a list of 7 birds in residence (that I knew about)
Common Redpoll
Lesser Redpoll
Linnet
Willow warbler
Wren
Sedge warbler
Robin
And in passing
Swallow
Oystercatcher
Curlew
It did not recognise the cheeping of the pheasant brood
It failed to identify a red kite from its call but when I described it, it got it right!
Described it? How do you describe a bird call to an AI app. Do you do it yourself as a sound, or do you write something down onomatopoeically?
.
(This may be obvious to birders, but I’m not one. )
Merlin is good with bird calls, song and even little mutters. It is just possible to fool it with mouth-sounds. I can ‘do’ a willow warbler. Most people can do a cuckoo or an owl.
But Merlin has a section that allows for a Step by Step iterative description of the BirdYouSee.
I recommend the Merlin App just for the calls and songs.
There is an AI ID Project running here https://www.ispotnature.org/communities/uk-and-ireland/view/project/865652/
Back to the original question…
I am thinking about it… I just read this which is apposite.
[quote=“Surreybirder, post:1, topic:1997”]
technology
[/quote].
.
The importance to us on ispot is at a different level & I’m thinking about that too.
Thanks
AI ID Apps aside. It seems quite important that WE should stay abreast of the way Technology is being integrated into most aspects of Natural History.
Should I devote a passage in the iProject to AI developments?https://www.ispotnature.org/communities/uk-and-ireland/view/project/865652/artificial-intelligence-id (Preview removed)
It could be useful but I think you would need to define some parameters - as you’ve written it, it would be a massive subject. Perhaps something like: ‘how AI is being applied/developed to help wildlife conservation’?
In response to your earlier point, Merlin is pretty good. It’s a help to me as my hearing isn’t good (despite digital hearing aids) and when I half-hear a bird and suspect e.g. a bullifinch, it’s great to be able to get it confirmed or otherwise. But it does sometimes get things wrong, so ‘buyer beware’.
Merlin told me I had a Tawny owl calling from just outside my window last night.
So, back to the original Question, here are my Idle Speculations on Automated Identification and other technology.
The availability of information « online » would likely be amazing to anyone whose experience had stopped with Interlibrary Loans requests. Many of us who have been around as the technology has become available have experienced daily digital technology wonders with occasional seismic changes.
.
There seems to be agreement here that FastCat is great with bird ID. At the other end … consider mosses. The naked eye can identify some species, a hand lens certainly comes in useful but a m’scope is often needed. And experience usually helps too.
But the current ID apps are not often right not even getting a bryophyte. I’ve noticed many cushion mosses seem to be Campylopus introflexus, an introduced species in UK and just one of many common mosses. Other mosses posted on ispot get AI IDs of grasses, Angiosperms, Pteridophytes. I have yet to see a correct one. Some way to go, then for moss AI ID.
For me the joy is using any of the available tools to arrive at an ID I can feel happy with. If other ispotters agree that’s further joy; if they disagree it’s educational.
Currently the digital technology is relying on data collected from the web. That data, for the most part, has been collected and labelled by a human, some of it many decades ago, some of which has been uploaded for us to use today.
So if the default is to move to reliance on AI apps will people like us, interested in knowing “what it is and why,” stop being interested? Probably not. But will there be others to replace us when, as is inevitable, we stop ( i.e die).
Will it happen that AI apps will use then information which has been determined by AI? Maybe it already does.
I’m in favour of any tool that helps me. Overall as usual, I think as Surreybirder said, for us “User Beware » is the way forward. Whether AI algorithms will do the same, who knows.
@Surreybirder Always, I wonder what it’s all about - where one needs to draw the line - then get back to thinking of this book The Tao of Pooh - Wikipedia
Let’s just enjoy nature - does it have to be super-scientific.
Thinking of how little children find nature so special.
I looked at observations of a number of Malvoideae species (Abutilon megapotamicum, Hibiscus syriacus, Malva sylvestris, etc.) on Pl@ntNet. There’s a significant number of incorrect identifications. I left feedback on those observations, but I’m not sure if anyone saw it.
Interesting you mention mosses as they are one of the groups that has shown a very large decline since 1970, details will be shown in upcoming state of nature report. When I saw that I wondered how much it was due to real decline and how much down to decline or at least change in recorders. Some of the data has fancy models to account for recording effort but not the actual recorders, were there more expert moss recorders in the past but now there are more people who can record some, especially the common, species.
This long preamble is basically agreeing that relatively easy things may get more records via AI but this may even lead to other bias in the biodiversity data.
I like JoCs summary - it fits well with my own Philosophy
Please do NOT see the project here https://www.ispotnature.org/communities/uk-and-ireland/view/project/865652/ as a Promotion for the Concept
And DO please look at some Obs in the Gallery. Perhaps, even, comment.
And please look at the way the ID developed here
https://www.ispotnature.org/communities/uk-and-ireland/view/observation/869251/
I think that we should have enjoyment of nature and be ‘scientific’. I’m not sure that ‘super-scientific’ means anything? For most of us it’s a hobby. And we should enjoy a hobby. Each of us may enjoy rather different aspects of wildlife. For example, I am not happy if I see a bird and don’t know what it is. Many taxa, of course, I just have to accept that I will not ever know what they are. But I still enjoy the ‘detective work’ of trying to ID some of them. If AI helps, I will use it. But as DejayM has repeatedly warned us, it can lead us astray! One of the worries about AI is that it does not necessarily test the validity of the data it is using - does it? So it could generate errors and thus replicate them. The example of the mosses (@JoC ) might be a case in point.
But I have no problem at all with people who enjoy looking at an environment without having any curiosity about its make-up, ecology, conservation status, risks and opportunities. And I’m sure that many such people feel that I’m highly eccentric - not many people wander around my neighbourhood with a sweep-net and binoculars!! (I do get stopped every now and again and asked what I’m doing - which can lead to interesting conversations. One young woman with two kids turned to them and said: “That’s a good idea. We should look for some wildlife.” I was pleased!!)
.
Who will publish this report?
Think it is RSPB that are the lead but involves many wildlife and research organisations.
the link is to a previous report, the current one not quite out yet.