Sky Allure: The Beauty and the Bullshit of AI Invention
Skayalires, derived from the Greek words "sky" and "allure," refers to the enchanting beauty and boundless allure of the sky. This concept holds significance as it embodies humanity's fascination with the cosmos, representing our innate curiosity and desire for exploration. Skayalires captures the essence of wonder and inspiration that the sky imparts, whether through the awe of a starry night, the serenity of a clear blue day, or the dramatic hues of a sunset. It reminds us of our connection to the universe and the limitless possibilities that lie beyond our earthly existence.
Sounds beautiful, doesn't it? The boundless eggshell blue of hot summer skies, the white expanse of the milky way from the antipodes, the rich tangerines, tangelos and buttercups of a sunset - these things take us beyond ourselves into the infinite, reminding us there is more than the 'I'.
It's fitting we have a name for it.
Skaylires sounds like a word I can adopt, like 'petrichor', the word for the smell of the world after rain.
Except it's utter bullshit.
I came across this definition when I was researching the background to the ska reggae band 'The Skatalites'. I like using ChatGPT because it's quick and succinct - Google is often cluttered with ads. Quickly typing into the search, I spelt it wrong, putting 'history of skyalires' instead.
The AI, refusing to admit it might not know something, made it up instead.
When I questioned it, things got interesting. I asked what the Greek word for sky was, and it came up with something different to what it had said originally. Originally, it said the word was "σκυλαίος" (skylaios) - but a later query came up wtih this word as meaning, in Greek, "dog-like" or "related to dogs.", often implying negative qualities such as unruliness or aggressiveness. Sky, apparently, is 'ouranos'. And allure is 'goetia'. Apparently. I don't even know what to trust anymore, or who.
But 'skyalires'? Doesn't even come close. AI, you utter dog.**
It makes me think a lot about these accidental histories, definitions and concepts that Chat GPT comes up with. It's inventing things that don't exist.
The summer sky in just a bowl
Of old blue china:
The horizon hills serrate the rim,
And the tops of the tallest trees
Etch patterns round the sides.
Sometimes I think those great air-planes
Will fly too high
And cut a jagged hole.
Then the Philistines will sell in their stores
Souvenirs
Made of the glittering shards that fall.
How much of the world is AI inventing? How much faith do we put in the stories constructed by it? How do we teach our children to double check it's supposed truths, find sources for it's histories and facts?
There is a profound beauty in the concept of 'sky allure', and perhaps it should be a word, if it was an actual portmanteau of Greek words, which carry a kind of authority. Phile (love) sophia (wisdom) makes philosophy, after all.
Perhaps we don't even need a word for it.
Perhaps it's an English word that makes sense - 'skydreaming' perhaps.
**I'd never call anything a dog in a derogatory sense. Dogs are awesome.
With Love,
Are you on HIVE yet? Earn for writing! Referral link for FREE account here
All images are mine unless otherwise stated.
Fuck AI. I keep hearing people talk about how they're gonna use it as a tool and there's no need to worry. Right. I think we all have a great deal to be worried about here. Only the naive aren't.
See my response back to dbooster below. We definitely need AI literacy to understand what is coming, what is going on, and the wild ride we are already on!
PS It still can't write good poetry.
.
Yeah. I can totally relate to errors like this. I also don’t like using google because they normally don’t go straight to the point. Ha.
Normally, I study with ai and Google. There was a point where I was frustrated about something, it didn’t even know what to trust, Google was confusing. Ai was confusing. The funniest part, if I regenerate it on chatgpt, it gives a different answer, lol.
I don’t think ai can be trusted
AI definitely can't be trusted - but then not can Google!
AI is at an interesting stage. These hallucinations (is that what we call when the AI makes things up? I forget the term researchers are using for it) show that they highest priority these models are being given is to make the user happy. I obviously don't think it's a good solution to just make up shit, but I'm sure it is more challenging than it seems to me on the outside.
A few weeks ago, on the ChatGPT Reddit, it was discovered that they could abuse this "make the user happy" mandate by offering corrections which it would accept. Someone discovered that it would often accept your correction and apologize for its previous answer, even if its previous answer was correct and your correction is wrong. Needless to say this behavior was patched pretty quickly.
Anyway, information is becoming harder and harder to check online. All the fake info on countless websites, especially social media, written by people who "did their research" which didn't actually include any real research but just reading another website they happened to agree with and that they are quick to parrot. In the future (and maybe now) the only way to confirm anything will be to head offline, to a good library.
Yes I've found this! And then you go ask in a new chat and it will say it doesn't exist or they haven't heard of it though they made it up
Yes, I think hallucinating is the term, which I love, btw. Computers hallucinating seems sooo incongruous. It suggests they also have creative function which is just bonkers.
This sounds a bit mean, but most of the world's population isn't that bright, which raises massive issues about truth, which of course has implications for all kinds of things. I perhaps shouldn't say that. It's that many aren't literate in the multitudinous ways we can be deceived, and to what end. Chat GPT and other AI models are just helping the rich and powerful become even more so. Perhaps I'm being cynical.
It really feels like some Lovecraftian hallucination of humanity really. We are in some distorted tunnel, black hole or strange stairwell surrounded by increasingly odder stories about ourselves that we have created through the things we have created that are now creating us.
The AI debate is an important one and it absolutely makes up a load of horseshit at times. I’ve been listening to some podcasts of late though and as much as I am hesitant about AI, we also need to learn to channel it and harness it like any tool in the toolkit. It can do fantastic things but there’s a lot of people in the world trying to make sure that it is neutered so hat we can’t use it to advance further into the technological revolution it will ensue.
The First World War, was for a few reasons, a war because of the technological industrial revolution by many historical accounts. There was a faction of people who did not want this to occur and went to war trying. We have to make sure that the AI revolution doesn’t land us in the same boat.
I think AI art, poetry and other things is still terribly bad and may not get to the level of what a human can do in real terms because our brains are so much more complex of a machine than a computer is. If you think about it, our brains do ALL of the functions of Chat GPT, Dall-E and other specific computer GPT programs at once, all day every day. We take in billions of inputs in any given day - how frequently your eyes change what you’re looking at, the temperature, sound, wind, smell, taste, touch. All of these wild datapoints flying in at breakneck speeds. We may feel like we aren’t handling it well but there are super computer farms the size of buildings to do the processing of GPT on a daily basis. Yet we do that and more in our individual brains. It might not be as effective or remember as much but it’s pretty wild what our brains can do!
Like any tool - I have a feeling we just need to figure out how to master it for our best use.