And, well:
What are we losing in this fire? Writing is not just about rendering thoughts through words in a certain style: analysis, literary fiction, storytelling. It is about the particular alchemy of a single individual drawing on their own unique profile to construct an idea. It is about the way their brain works, the quirks they have picked up along the way, their politics, their history, their relationships, the very way they see the world.
Awww, Gawd. No, don’t lassie. Writin’ ’bout ‘ritin’ is very boring.
I see nobody is commenting on this one. Not much you can say about a pile of pretentious crap.
“And every time I see a story of a journalist caught out by fake AI quotes during research, I cross myself – there but for the grace of God go I.”
If you’re putting in fake AI quotes, it means you’re writing low-grade regurgitation. You don’t understand the subject in any depth. But avoiding AI because of that risk just means that you’re going to get other trash, like all the writers who don’t understand what a mineral reserve is.
The fun thing is that ChatGPT replaces most columnists. It regurgitates the opinions of a group, which is what most of the Times, Telegraph and Guardian are now. It’s bollocks but so are they.
An interesting point is that ChatGPT (and Claude, Gemini etc) all do “know” what a mineral reserve is. Or at least can regurgitate some definitional stuff and get the factual details right. I don’t know if you tell it to write a piece of pure rhetoric it will stray into using the word wrongly, since that’s how it’s used in the rhetoric it’s been trained on.
I always advise caution about using AI for anything “factual” but (often deliberately) ignorant or under-informed humans are quite capable of writing their own nonsense, and AI isn’t always a downgrade on factual accuracy. Just a question of how much veracity matters to you. If you’re using AI for direct quotes then sadly it’s a sign that you didn’t care at all.
It’s fine for where the answer doesn’t matter much. No-one uses the Guardian for anything serious. It’s a form of entertainment.
Sadly I meet far too many people who believe what they read in the Guardian. Even if it’s in an opinion piece or editorial.
You think you’ll be careful, you will double check everything. And then your book comes out and it appears that it includes more than a half dozen misattributed or fake quotes.
How does one ‘double-check everything’ and yet fail to track down quotes?
“Writin’ ’bout ‘ritin’ is very boring.”
Shakespeare wrote some good stuff about acting. But he was a genius.
ME, ME, MEEEE!
“the particular alchemy of a single individual drawing on their own unique profile”
When the profiles of all the writers at a newspaper are so similar, it turns the newspaper into pap.
Not every story is worth hearing.*
*Most aren’t.
Have you noticed that, when using AI, the computer regularly asks the human operator to confirm they are not a robot. Will this be the future of the human race?
When they ask you to confirm you ARE a robot.
It’s not real AI if it can’t prove it’s human. How hard is it to tell which frames have a motorcycle in them?
Try it withThe Terminator.
“Are you a robot ?”
“I’ll be back…”
There is a widespread error in thinking about AI; people assume it’s a digital research assistant. So when it gets things ‘wrong’ they call it ‘dreaming’ or ‘hallucinating’.
But AI’s, at least the LLM AI’s, are not truth engines; they are pattern-matching engines. If you ask it for a thing, it searches what it knows about that thing and generates an answer that matches the pattern of that thing.
For instance, a few years ago when ChatGPT was everywhere in the news, I asked it for a potted biography of our own dear Timmie. I figured I’d been reading his stuff long enough to have some idea of what he was about.
The result was a perfectly cromulant-looking three-paragraph potted bio. The first paragrah was about Timmie himself, the second a list of his publications, and the third his professional associations.
The publications were easy enough to fact-check. Three books: the first turned out to be genuine, but the second was actually written by someone else and the third didn’t exist at all. The latter two, although misattributed or non-existant, nevertheless had titles that sounded like something Timmie would have written.
On it’s own terms, ChatGPT succeeded. I asked for a potted bio, it gave me a potted bio. Regardless of some of the ‘facts’ being wrong, it still matched the pattern of potted bios.
That’s why they cannot be relied on to do research. It’s not what they were designed to do.
I just did that – for research processes – with the Google AI. Has a commercial published book as being through ASI. More work needed, obvs.
I just tried myself, and it came out almost complete, accurate (I think) and entirely positive.
Hopefully that means it doesn’t read Murphy or other left-wing critics.
Actually it was more accurate than I’d have been about myself; I was about to criticise it for claiming I’d done something that I hadn’t, but then remembered that I had.
Inspired by your tale I have just asked Google’s AI to list my research papers. It mentioned a dozen – that must be wrong by an order of magnitude.
When I retired I shredded my CV so I can’t tell you how many papers I have published in all but I’m pretty confident that a dozen is fewer than I published after retiring! So the bloody thing has ignored forty years worth of output.
Or classed them as “derivative works”….
They also tend to miss the “**to whom correspondence shoud be directed” taglines, indicating the warm body who *really* did the grunt work, and is generally listed well behind even the secretary of the Head of Faculty…
I never asked Google’s AI to list my research papers (none of which were university standard ones, just for my professional newsletter/magazine) but I did once ask it for articles about me (just because you’re paranoid that doesn’t mean they’re not out to get you so I occasionally act paranoid in case) and found that it categorised me as two separate people (with a footnote that my great-grandfather had the same name but totally ignoring the existing of my late cousin and one of my great-uncles, both of whom were also called John).
I feel that I am allowed to call it “Artificial Stupidity”.
I discovered I saved the world.
Tangent: I have become well skilled at avoiding clickbait [spell checker accepted real word] on youtube.
AI, clickbait . . . we are approaching 1900. You can’t believe anything you see or hear.
Which Gamecock considers the best state. Always on the verge of anarchy. Government can’t help you . . . as it should be. Not their job, mon.
I tried to read this nonsense and had to quit part-way through.
Writing started as a means to transfer information (even if it was only about who owed how much in tax to the ruler). Nesrine has her own opinions and wishes to rewrite all of human history to fit her personal diatribe.
There is truth in history, anthropology and archaeology, on the other side there is Guardianista propaganda – never the twain shall meet
Yeah, but it pretty quickly descended into propaganda.
Pharaoh Ramases II claimed victory in the Battle of Kadesh (1274 BC?), when in reality he barely escaped with his life and lost the territory he was fighting over.
Nesrine is, sadly, just the latest example of a long tradition.