Tag: Artificial intelligence

Aaaargh! Spring – please slow down, just a bit!

Clockwise from top left- the canalside view with Pen y Fan in the background; then Common Dog Violet, Wood Sorrel, Maidenhair Spleenwort, Ramsons, Greater Stitchwort, Reflexed Stonecrop and what I think must be bracken growing very close to the water. There were many more – too many to list without annoying Madame!

The rewards of Spring are everywhere at the moment, notwithstanding the cold nights which are keeping our tender plants blocking the hallway as they harden off. There’s so much going on I hardly know where to begin. On the allotment – after the usual despairing survey of the weeds, the waterlogged ground and the mounting sense that nothing good will ever come of it; we got our heads down two or three weeks ago and felt instantly better. I’m a bit suspicious of the received wisdom that gardening is good for the soul. Couch grass and Bindweed could test the patience of a saint and I’m certainly not one of them. At the weekend while Madame sowed, I finally cleared the asparagus bed which had been on probation for ages and we knew it had to go because it was too far down in the frost pocket on our sloping site plus the asparagus had been weakened by repeated invasions of Asparagus beetle and couch grass from the unattended plot next door. Four barrow loads of weeds and feeble/floppy/extinct roots later I had a backache worthy of a third rate wrestler but a decent empty raised bed much enriched by previous additions of seaweed, compost and sand and with around 18″ depth of topsoil. We’ll grow carrots there this season. We’ve thrown so much money at the asparagus bed over the years, we could probably afford to buy fifty bundles of Chinn’s finest English and still be in pocket.

I was greatly assisted by two almost hand tame Robins who were obviously feeding chicks. Between them they took away many dozens of larvae, centipedes and other insects. Interestingly they weren’t very interested in worms; certainly not as keen as a blackbird would be. I was dazzled by their capacity to hold two wriggling bugs in their beaks and still pick up a third without dropping the first two; they were far better pest controllers than any chemical insecticide.

Inside the polytunnel the experimental crop of broad beans is thriving in the absence of any really hot weather, and the strawberries, all taken from runners in late summer, are flowering and setting fruit. Of course this means we’re already watering inside the tunnel; but the 12V water pump we bought last summer has already showed its worth and helped us avoid carrying heavy watering cans back and forth.

The photos above were all taken on a short trip to the Monmouth and Brecon canal near Brecon during the week. It was here, many years ago, that I saw my first Kingfisher – so beautiful in the sunshine that I thought I was hallucinating. We had hoped to go for a pub lunch with our friends who keep a smallholding almost 100o feet up on the hill, but they were in the middle of lambing so we had a picnic lunch there while they went outside every twenty minutes or so to keep an eye on a ewe in the midst of a long and difficult lambing. Fortunately the ewe and her twin lambs all made it through, although I think that will be her last time. Farming can be heartbreaking as well as hard work.

“A difficulty is a light, an insurmountable difficulty is a sun” – Paul Valéry

Is it too perverse to say that I love naming plants? and the harder they are the greater the reward when I finally get there. I’m exhilarated by the explosion of plants in the spring and early summer, and it’s agony having go forego plant hunting for allotment duties, but there’s no alternative so we just get on with it. I was pondering where this love of plants came from, and during one of my regular 2.00am wakeful sessions – it happens a lot – it occurred to me that I owe a huge amount to Henry Williamson (I’ll come to the reservations in a moment). Of course I read Tarka the Otter and the other nature books, but I also ploughed my way through four volumes of “The Flax of Dream” and fifteen volumes of “The Chronicle of Ancient Sunlight”. I was probably the only person who ever ordered them all up from Bristol Central Library in many years. Above all, I loved Williamson’s ability to describe wild plants in their landscapes; their names – English names – embedded themselves in my imagination and make the discovery of a plant in a hedgerow into a celebratory event, even fifty years later. Latin names and taxonomical exactitude; whilst essential for research, are feeble by comparison with the poetry of use and history.

But one of the greatest sadnesses of my life has been the discovery that so many of my literary and artistic heroes dabbled with and even collaborated in extreme right politics during the nineteen thirties and forties. TS Eliot, Ezra Pound, Wyndham Lewis were all seduced by the big lie. Discovering that Williamson was an admirer of Adolph Hitler irrevocably shut down my relationship with him, and I’ve never read a page written by him since; but the influence of his natural history writing still remains – it’s just forever tainted by the association.

Anyway, turning with relief to spring again, some flowers that you’d think were easy to identify – are more than a bit fiendish; not least the violet which comes with seven close cousins six of which you could easily bump into in the South West. No alternative, then, but to turn to the books or the apps. But when it comes to fiendishness, nothing comes close to ferns and for me, at the very beginning of this love affair, a chance encounter across a crowded room can lead to hours of agonising – just like the real thing (I’m told!).

As you’ll know if you’re a regular reader, I’m exploring the dizzy world of Artificial Intelligence in wildlife apps, and particularly in identifying plants. If you’re fortunate enough to know what a data point is – I wasn’t – it’s a single unit/dollop of data. If you’re still attached to pencils and paper, your notebook might contain a few hundred data points. A field guide could hold tens of thousands, but AI robots, though are voracious readers and can consume and store billions of them. Not only that, they can index and arrange them in pretty much any way you like.

So before we all get carried away by the idea of wildlife AI apps remember that the whole industry is based on text. I’ve been playing with Google Gemini but other flavours are available. These text engines can be based on anything up to (I believe) 8 billion data points – that’s a lot of text and a huge fund of examples to work from. The existing wildlife apps are still wallowing in the relative shallows and so they can be unreliable at the moment. I had three obviously wrong fern identifications (back to the books) while we were up on the Mon and Brec. They’ll get better very soon I’m sure. The Botanical Society of Britain and Ireland have access to over 50 million full records, and Kew Gardens are digitizing their entire herbarium records. iNaturalist also has something huge like 50 million although not all of them are verified, but even if all these records are verified and scanned in – which would be a huge volunteer operation – they would still be far fewer in number than the mighty text warehouses. Machine learning can achieve seemingly miraculous results but I don’t think we’ll be making human identification redundant any time soon, so don’t throw those field guides away!

Just to finish, though, I thought I’d wile away an hour asking Google Gemini to do some silly things for me. Question one – what are the distinguishing features of Dryopteris ferns? after ten seconds a very sensible answer. Then I asked it to write a sonnet on the subject of dust and once again a technically perfect but aesthetically clunky, sub Tennysonian sonnet emerged. Then finally I asked if it would re-write one of these posts in the style of Phillip Marlowe. The result was hilariously funny but quite unprintable here, being vulgar, deeply sexist and full of bad language.

Thank your lucky stars it’s just me writing this one!

C.A.T. sessions (close attention to text)

The battery charger in the campervan.

I was going to call this post “punkt, point, period, full stop”. But decided to call it C.A.T. in honour of a very fine teacher I once had. His name was Canon David Isitt and he was the joint Principal of the Bristol and Gloucester School of Ministry and I was there to teach some sessions on communications but when David was around I was always a student. On training weekends when the weather was good enough we would sit in a circle on the grass outside the retreat house as if he were Socrates and we were the oikos and would spend a couple of hours distilling the myriad possible meanings of a verse or two from the Bible. The relevance of this charming memory is that today I solved a longstanding problem (my idiocy) by paying minutely close attention to a single Google generated password. For over a year I’ve been beating my head on a wall over my inability to access the OS map app on my phone because it kept rejecting the password. Today in a moment of sub religious ecstasy I noticed a single full stop at the end of it which my smartass phone had never copied because it thought it signified the end of a sentence. I added the dot and the app sprang into life as if by a miracle.

As of now I’m handing in my tools

I wish I’d paid the same attention last September when I replaced the £250 charger unit in the campervan by taking a photo of the original one (see above and despair) and replacing it wire by wire with the new one. That never worked either and in all our subsequent campings we would sit in the van with head torches watching the batteries go flat. Previously I’d done the same with the gas jets on the three way fridge with the same result – didn’t work – which was the point of our hideously expensive trip to the workshop on Dartmoor a few weeks ago when my handiwork was repaired but mostly replaced by some great guys who actually knew what they were doing. So the batteries now charge, the fridge works, the sink that always leaked after I replaced the plug has been renewed, and the satellite dish which fell off the roof on the way back from Brecon has been replaced by a new miFi router which works like a dream. As of now I am handing in my tools.

If you are sensing a new resolution and vigour in my mood, it’s because I’ve finally found out what level of beta blockers I need to take without becoming a zombie or listening to my heart thrashing itself to death. I’ve found the sweet spot, I think, and I feel better than I have since last September. Perhaps there’s a clue in the dates! Never undertake electrical repairs when your brain isn’t working.

So today we went down to the van and tested the new systems and they all worked. We have booked a couple of short breaks in some favourite places with good botany potential, and I’ve spent the past two weeks designing the best possible workflow for submitting records – which included learning an entirely new application; iRecord – if you’re at all interested. Having offered to do a talk on useful apps to the Natural History Society, I wandered off into the impenetrable scrub of phone apps for identifying and recording wildlife. As ever, I entered the thicket with the fullest knowledge of what I might find there and promptly discovered that my understanding was mostly cobblers. I began by being quite certain that iNaturalist was by far the best recording app, and then after a few small irritations I fired off some emails to naturalist friends who mostly thought there were big problems with it – not least the verification process which, it seems, might allow three or four people with even less knowledge than me, to tick a box and elevate my incorrect identification to “research grade”; good for my ego but very bad for science. It seems I was conflating easier to use, with best; an error from which none of us are immune.

So, if not iNaturalist; which? Most of my enquiries trickled into the sand when it came to the big beasts at County or National level. Natural history recording, it becomes clear, is like Italy before unification; awash with fiercely defended kingdoms about the size of your average hamlet, and one particular bone of contention is the use of artificial intelligence – please feel free to rub that clove of garlic on your laptop.

The problem with identification is getting worse and not better. DNA analysis has thrown a boulder into the pond – I’m a stranger to any deep understanding of the subject, but the names and familial connections of living things seem to change all the time. The identification of plants by their appearance – morphology in science speak – is no longer as valuable as it once was; and that pulls the rug out from under one of the great strengths of AI – its ability to scan tens of thousands of photographs in less time than it takes you to find your notebook and come up with a plausible ID and name. Like stocks and shares, identifications can also decrease in value. Most of my learned friends prefer to lean on their many years of experience rather than trust a phone app, and if I had that much experience I’d probably feel the same way.

But what about those of us who love to look for plants. birds, mammals, fungi and all the rest – and regularly get stuck in an identification but would still like to make our contribution to the records in this era of environmental destruction. Getting your records confirmed can take weeks and even months but it’s imperative that records are checked by the best available experts before marshes are flooded and rivers diverted to save things that were never really there. Sometimes it’s best to give way and beg forgiveness when your record is rejected but sometimes – more rarely perhaps – it’s worth backing your hunch and asking for a second or even third opinion. So my money, at present, is on iRecord which uses the same family of software as iNaturalist – designed by an outfit called Indicia; hence the similarity of web design. In the iRecord app, you have to enter some kind of identification before you can add photos – then AI will tell you in percentage terms whether you nailed it or failed it. The saved data is made available to hundreds of skilled volunteers who can either agree or disagree with the machine intelligence. The software puts human intelligence firmly in charge. The other great advantage of the programme is that when all’s settled, the record is shared with all the appropriate local and national recording schemes (the warring states?). The data from iNaturalist is also now passed to the British databases but has to go through the same verification process as it would coming off my phone – which is a tremendous waste of volunteer effort.

What’s in it for us apprentices? Well I’ve found that the process of getting some sort of name necessarily involves turning to the books; learning families, environments, seasons and relationships through a process of reinforcement – good educational practice – we all get quicker and more accurate; and therefore more useful in the fight against environmental destruction.

I’ve had a comical vision of future field trips in which participants carry enormous pieces of computer and electronic kit across the fields on converted trolleys with old pram wheels (we’re like that) and stand in silent circles around wiFi aerials and routers gazing at our phone screens in silence. Actually, that’s not funny at all, and in any case hand-held DNA sequencers can already be had at huge cost; sequencers which rely on pre-existing DNA databanks, which can be searched using (you’ve got it) AI. I much prefer crawling around on my knees in the rain and mud. AI is alright for some jobs, but it will never have those moments of left-field inspiration that human minds are so good at. I shared that thought over coffee this morning with (name drop alert) our neighbour Prof Charlie Stirton who came up with exactly the same proposition. High fives then!

So in the next three weeks we’re off in our newly and fully functional campervan and spending a few nights on the beloved Mendips and in the Brecon Beacons – Bannau Brycheiniog means the peaks of Brychan’s kingdom – and later a grand tour of Dartmoor, then West and East Cornwall to catch the spring flushes of flowers while I test a few more phone apps and websites for my talk. Life is good!

A creaking gate lasts longest!

Male fern – Dryopteris filix-mas – I think; with Hart’s Tongue which at least I’m sure about!

I blame Helena, our VC6 (North Somerset) County Recorder, for getting me interested in ferns. Madame and I had joined a field trip to the Mendips, and in particular to a nature reserve elegantly named “GB Gruffy” The day is etched in my memory for two reasons; firstly because Helena spotted and named an unusual fern nestling six feet down a gated mineshaft. At the time my knowledge of ferns was confined to Bracken and Hart’s Tongue and I wasn’t even sure about bracken, so I was filled with admiration for her expertise. The other reason for remembering was that somehow I lost a rather expensive telescopic lens whilst yomping across the tussocky clumps in a deep bog.

Ferns have been around an awfully long time – around 300 million years – so they can certainly claim longevity in addition to a complicated sex life and the gift of occasionally doubling up on their chromosomes. They are – it’s true – very challenging to identify, or at least some of them are, and so they’re also fatally attractive to propeller heads like me. So after my brief excursion back to my old day job which really did stir up the silt of memories at the bottom of my pond, Madame made sure that our time was filled with anything that didn’t involve me wearing a frock. Distraction therapy, you might say. So we went up to Mendip to hunt for fungi – rather unsuccessfully; then we went to a fine lecture on bees – but not honey bees – and I found myself volunteering to lead a field trip in the spring, the thought of which is terrifying because I’ll be with a couple of co-leaders, a birder and an entomologist who really know what they’re up to. Imposter syndrome is a painful business! We drove back up to Priddy in the campervan for a couple of nights but the trip was overshadowed by heavy rain and thick fog, so we came home a day early. Since then we’ve been vaccinated for flu and Covid and I’ve had my new drug regime finalized. There’s nothing fatal wrong with me except worn out joints and an over excitable heart which requires that I take medicines with nasty side effects and which take weeks to bed down. My only concession to all this is to wear mittens a lot of the time because I now have Raynauds and my fingers get painful and stiff. I’m not quite 300 million years old but it occasionally feels like it, and so I’ve become a bit of a creaking gate.

Now prepare yourself for a true stinker of a link because a real creaking gate featured in yesterday’s walk. The sun was shining and Madame, continuing her campaign of loving distraction, took us off to Newton Park for a stroll around the lake so I could look for ferns and try out three new ID apps on my phone. This is going to be the subject of my talk next spring – phone apps and AI and their strengths and weaknesses.

Newton St Loe is a place that seems to be wholly owned by the Duchy of Cornwall and so it’s a picture perfect village where even the no parking signs are made from cast iron. We parked the car as far from the signs as we could and bumped into a party of about a dozen people led by a man wearing a Viyella shirt with well pressed trousers and gleaming brown shoes. I concluded that he was a land agent or some such because they were all laughing at his jokes. We joined them as we walked up the road towards the church and unwittingly divided them into two groups. As we left the churchyard I became fascinated by the creaking of the gate because it sounded three distinct notes and so they waited a bit impatiently as I swung it to and fro and even sang along with it. I love the sounds that gates make. There’s a broken gate made from tubular steel on one of our favourite walks where we camp at St Davids. It sings sweetly like a flute and depending on the wind strength will even fluctuate over several harmonics. S

]Natural sounds are so important. Later as we sat alongside the lake I was trying out a birding app called Merlin – which is amazingly accurate. There were few birds that could compete with the sound of the wind in the still fully leaved trees, but crows, jackdaws, coot and mallard were all calling. Aside from that our best sighting was a hornet which dashed for cover among some laurels, and I found lots of Male ferns, which isn’t surprising because they’re ubiquitous in the UK. On the other hand I do at least know now what they’re called and – being a bit of a creaking gate myself – I could just have 299 million years left to learn the rest of them. But I’m not holding my breath.

Postscript

Having written this piece, I realized in the middle of the night that with a little bit of detective work I could probably find the name of the fern that Helena spotted – apart from asking her, that is. So there’s a very useful document from the British Geological Survey which I often refer to, called the Biodiversity of Western Mendip which covers most of my favourite places. Turning to the section called GB Gruffy Site I discovered that a moderately unusual fern called the Brittle Bladder fern, Cystopteris fragilis occurs on that site but just to double check I went to the BSBI Atlas 2020 website and searched for it. One of the key tools for finding plants is to know their habitat and so when I read that this fern is most often found growing in the semi darkness of cave entrances and mineshafts and then found a confirmatory 2 KM dark square with the site in the middle I was delighted. Even more delightful was the news that one of my long-term bucket list plants – the Spring Sandwort, Minuartia verna also grows nearby. All I need to do is wait ’till next spring!