The Cringe/Hate Thread

[Which Tyler]

The amount of times I hear, there’s » before plurals!
I hear it most days on BBC Radio4 and I shout at the radio, “There are, ffs!”, and as I work as an English teacher over here, it drives me crazy, as do the increasingly number of Americanisms that are prevalent in our language today, as mentioned by Which Tyler in the previous post in which The Guardian is a serial offender!

Would of

[Donny osmond]

The amount of times I hear, there’s » before plurals!
I hear it most days on BBC Radio4 and I shout at the radio, “There are, ffs!”, and as I work as an English teacher over here, it drives me crazy, as do the increasingly number of Americanisms that are prevalent in our language today, as mentioned by Which Tyler in the previous post in which The Guardian is a serial offender!

Would of

Sayings like "I've went ... " or "I've ran ...." are starting to annoy me, but they are, in my experience, mostly Glasgow sayings. However, I've noticed a few BBC types who should really know better starting to use the same incorrect English with sat versus sitting. As in, "I was sat in the fannies room ay Glasto". Big sigh.

[francoisfou]

The amount of times I hear, there’s » before plurals!
I hear it most days on BBC Radio4 and I shout at the radio, “There are, ffs!”, and as I work as an English teacher over here, it drives me crazy, as do the increasingly number of Americanisms that are prevalent in our language today, as mentioned by Which Tyler in the previous post in which The Guardian is a serial offender!

Would of

YES! That’s another that drives me up the wall!

And ……. the regular use of « less » before a plural!

[Galfon]

'Reimagine' - really doesn't look right on paper..
Seems like revamp / transform or even recreate / refresh don't describe the genius work going on sufficiently any more.
Can't imagine why.

[Puja]

AI.

Yes, I know, that's a fairly basic answer - it's accelerating the destruction planet with the exorbitant use of energy, water, and rare earth minerals, it's solidifying further the tech oligarchy that's driving gross inequality and fascism, it's corrupting the greatest respository of human knowledge that has ever been created and rendering large chunks of it functionally unusable, it steals from artists of all stripes and devalues human invention and creativity, it's making the populace as a whole dumber and degrading education, and it's making it childishly easy to believably lie and cheat which benefits politicans and con-artists alike.

But aside from those minor things, the worst thing about AI is that it writes like me.

Em-dashes, over-complex sentence structures, lists/bullet-points, circumlocution, using abstruse words - these are all Puja classic moves*, but they're also things which people say to watch out for when trying to tell if text is AI-generated. I don't want to change my writing style, but I am living in fear of the day that someone looks at some screed that I've spent 30 minutes of my time carefully crafting (usually in lieu of doing my actual paid work) and thinks that I am the kind of awful person that lets a computer do my thinking for me.

Puja


*To the extent that this very post contains probably a dozen examples and not a one of them intentionally placed there for the bit.

[morepork]

AI.

Yes, I know, that's a fairly basic answer - it's accelerating the destruction planet with the exorbitant use of energy, water, and rare earth minerals, it's solidifying further the tech oligarchy that's driving gross inequality and fascism, it's corrupting the greatest respository of human knowledge that has ever been created and rendering large chunks of it functionally unusable, it steals from artists of all stripes and devalues human invention and creativity, it's making the populace as a whole dumber and degrading education, and it's making it childishly easy to believably lie and cheat which benefits politicans and con-artists alike.

But aside from those minor things, the worst thing about AI is that it writes like me.

Em-dashes, over-complex sentence structures, lists/bullet-points, circumlocution, using abstruse words - these are all Puja classic moves*, but they're also things which people say to watch out for when trying to tell if text is AI-generated. I don't want to change my writing style, but I am living in fear of the day that someone looks at some screed that I've spent 30 minutes of my time carefully crafting (usually in lieu of doing my actual paid work) and thinks that I am the kind of awful person that lets a computer do my thinking for me.

Puja


*To the extent that this very post contains probably a dozen examples and not a one of them intentionally placed there for the bit.

"in lieu of " is one of those...

[Puja]

AI.

Yes, I know, that's a fairly basic answer - it's accelerating the destruction planet with the exorbitant use of energy, water, and rare earth minerals, it's solidifying further the tech oligarchy that's driving gross inequality and fascism, it's corrupting the greatest respository of human knowledge that has ever been created and rendering large chunks of it functionally unusable, it steals from artists of all stripes and devalues human invention and creativity, it's making the populace as a whole dumber and degrading education, and it's making it childishly easy to believably lie and cheat which benefits politicans and con-artists alike.

But aside from those minor things, the worst thing about AI is that it writes like me.

Em-dashes, over-complex sentence structures, lists/bullet-points, circumlocution, using abstruse words - these are all Puja classic moves*, but they're also things which people say to watch out for when trying to tell if text is AI-generated. I don't want to change my writing style, but I am living in fear of the day that someone looks at some screed that I've spent 30 minutes of my time carefully crafting (usually in lieu of doing my actual paid work) and thinks that I am the kind of awful person that lets a computer do my thinking for me.

Puja


*To the extent that this very post contains probably a dozen examples and not a one of them intentionally placed there for the bit.

"in lieu of " is one of those...

That's a very interesting suggestion morepork. Let's delve into that.

[morepork]

Oh you terrible cynic. I’ve been fingered for that phrase recently, and not in a good way.

Incidentally, AI has made complete donkey cock of scientific references in “edits” from colleagues. Authors, dates of publications and titles all an incorrect application of fact. AI is not just fucking up lexicon, it is sodomising peer review, again in a bad way.

[Which Tyler]

Oh you terrible cynic. I’ve been fingered for that phrase recently, and not in a good way.

Incidentally, AI has made complete donkey cock of scientific references in “edits” from colleagues. Authors, dates of publications and titles all an incorrect application of fact. AI is not just fucking up lexicon, it is sodomising peer review, again in a bad way.

It's been explained to me (I don't know how accurately) that AI's job isn't to answer the question asked, but to show you what an answer to the question would look like.

It doesn't go scraping academic journals for conclusions, abstracts and information, it returns answers that look as if it might have done so. So it will give a plausible looking reference to something that looks like a journal, written by someone who's name looks like an academic, and with a title that looks like one that that journal might publish.

[Puja]

Oh you terrible cynic. I’ve been fingered for that phrase recently, and not in a good way.

Incidentally, AI has made complete donkey cock of scientific references in “edits” from colleagues. Authors, dates of publications and titles all an incorrect application of fact. AI is not just fucking up lexicon, it is sodomising peer review, again in a bad way.

It's been explained to me (I don't know how accurately) that AI's job isn't to answer the question asked, but to show you what an answer to the question would look like.

It doesn't go scraping academic journals for conclusions, abstracts and information, it returns answers that look as if it might have done so. So it will give a plausible looking reference to something that looks like a journal, written by someone who's name looks like an academic, and with a title that looks like one that that journal might publish.

This is exactly correct. AI doesn't **know** anything, it doesn't actually understand the question or prompt that you ask - what it does (oversimplified) is create an answer based on what its dataset says answers to that kind of question look like. It's very complicated and very clever and, when used correctly, can be incredibly useful.

However, if you ask it to write you a scientific paper, it won't include references because it knows what a reference is. It definitely won't include them because it's used information from other papers and is correctly citing them. What it will do is include bits of text in the format "(Morrison et al., 2019)" after certain sentences, because its dataset says that other scientific papers also include bits of text that look like that, so the correct answer to "Write a scientific paper for me" will include text in that format.

The problem is the branding - it's not Artificial Intelligence; it's a Large Language Model that has been trained with the goal of producing correct-looking writing, not correct information. There are a lot of occasions where it *will* produce a correct answer, because producing a plausible answer from a dataset of correct information will often do that, but that is chance, rather than anything being understood or any intelligence being involved. Yet it is advertised as "AI" that "understands" and "can answer questions" and people believe that it is a general purpose answer machine that is the spiritual successor to googling something, when all it is is a text-generation system.

It's a shame, cause I suspect that you probably could train an 'AI' to look up information and write a scientific paper, although probably only if you started from scratch with that sole ambition and spent a lot of time and money on it.

Puja

[J Dory]

AI isn’t perfect, but some of the complaints here make it sound like it’s plotting the fall of civilisation from a server rack.

Energy use? So does Netflix, crypto, and everyone doomscrolling at 2am. Funny how only AI gets blamed for melting the ice caps.

Big Tech control? Meanwhile half the planet is running open-source models on a laptop. Not exactly Skynet.

“It’s ruining knowledge!” No, people copy-pasting garbage without checking are ruining knowledge. AI is a tool, not a babysitter.

“It steals from artists!” If learning from existing work is theft, every artist on earth is guilty.

“It makes people dumb!” People were doing just fine on that front long before chatbots arrived.

Hallucinated citations? Ask it to invent a paper, it invents a paper. That’s not a crisis, that’s cause-and-effect.

AI isn’t the devil. It’s just another tool—brilliant when used properly, disastrous when used by people who shouldn’t be allowed near sharp objects or “Generate” buttons.

So says chatGPT with a couple of shorter and more sarcastic please iterations. Little more work and it could be MP

[Puja]

Energy use? So does Netflix, crypto, and everyone doomscrolling at 2am. Funny how only AI gets blamed for melting the ice caps.

On the one hand, I do also blame crypto, especially bullshit crypto like NFTs. However, this is incredibly specious - AI's power requirements are several orders of magnitude higher than "Netflix, crypto, and everyone doomscrolling", beause of the literal trillions of calculations needed to train and retrain the models. At present, AI-specific servers make up 60% of the energy and computational use of US data-centres and it's estimated that the requirements will double by 2028. By that estimate and my calculations, AI will be producing an additional 120 million metric tons of CO2 each year (and that's assuming the additional power comes from the same mix as currently, which is unlikely considering fossil fuel plants are easiest to build or ramp up to meet a massive new demand over the next 2 years).

Big Tech control? Meanwhile half the planet is running open-source models on a laptop. Not exactly Skynet.

And the half that's not running open-source models on a laptop? The half that you and I both live in?

“It’s ruining knowledge!” No, people copy-pasting garbage without checking are ruining knowledge. AI is a tool, not a babysitter.

It is a tool that's being advertised and marketed as a universal panacea and it is unsurprising that non-experts are trying to use it as such. Slop is starting to outweigh actual information on the internet, because it's easier and faster to make. It used to be if you needed a photo reference for something, you could go on pinterest and find a photo of an actual physical object. Very difficult to do that now, because it's overwhelmed with AI images.

“It steals from artists!” If learning from existing work is theft, every artist on earth is guilty.

"Learning" is the key word here. Someone that learns from existing work and creates something new using their own skills, mixing in what's inside them, is an artist. AI doesn't have anything inside it other than what other people have already made. It can't create anything new; it regurgitates from what it was trained on.

Let's also be clear, 99% of the time, the datasets that LLMs are trained on is stolen material. The artists and authors did not give any permission for their work to be used or reused and it is clear copyright infringement that is allowed to stand because of the closeness of techbros to governments and the resulting impossibility of getting a successful lawsuit.

“It makes people dumb!” People were doing just fine on that front long before chatbots arrived.

Always true, but before chatbots, people couldn't cheat their way through school and university. People couldn't ask a machine to do their thinking for them. You don't think that'll have an effect?

Hallucinated citations? Ask it to invent a paper, it invents a paper. That’s not a crisis, that’s cause-and-effect.

Again, back to the way it is advertised as "Artificial Intelligence" not "Artificial Plausible Writing Inventor".

AI isn’t the devil. It’s just another tool—brilliant when used properly, disastrous when used by people who shouldn’t be allowed near sharp objects or “Generate” buttons.

Good thing that it's not a massively powerful tool that's being marketed to all and sundry then and is only available for informed and educated users.

Puja

[Which Tyler]

Snip

[Mikey Brown]

AI isn’t perfect, but some of the complaints here make it sound like it’s plotting the fall of civilisation from a server rack.

Energy use? So does Netflix, crypto, and everyone doomscrolling at 2am. Funny how only AI gets blamed for melting the ice caps.

Big Tech control? Meanwhile half the planet is running open-source models on a laptop. Not exactly Skynet.

“It’s ruining knowledge!” No, people copy-pasting garbage without checking are ruining knowledge. AI is a tool, not a babysitter.

“It steals from artists!” If learning from existing work is theft, every artist on earth is guilty.

“It makes people dumb!” People were doing just fine on that front long before chatbots arrived.

Hallucinated citations? Ask it to invent a paper, it invents a paper. That’s not a crisis, that’s cause-and-effect.

AI isn’t the devil. It’s just another tool—brilliant when used properly, disastrous when used by people who shouldn’t be allowed near sharp objects or “Generate” buttons.

So says chatGPT with a couple of shorter and more sarcastic please iterations. Little more work and it could be MP

I get your point on most of those, but it’s a bit like the gun debate. Guns don’t kill people on their own, but we’ve seen the effects of widespread use by idiots and how quickly it gets out of control.

There are some fantastic uses for AI, but a lot of the examples above are happening whether it’s user error or not. It’s not plotting to destroy the world, but it could play a mighty big part through pure ignorance, and greed/manipulation of those behind it.

[J Dory]

Sorry, I thought I made that obvious. None of the above was me, it was ChatGPT after I fed it one page of this thread and asked it to prepare a response in defense of AI.

[Puja]

Sorry, I thought I made that obvious. None of the above was me, it was ChatGPT after I fed it one page of this thread and asked it to prepare a response in defense of AI.

Hah! No, that wasn't clear - was about to send you for an HIA!

Puja