To be utterly sick of AI already

How did you do this??

I was just reading about the French woman who has been scammed out of £1million after the scammer used AI video to pretend he was Brad Pitt and convinced the woman she was in a relationship with Brad.

It's got a massive carbon footprint too
Ah just seen the post above!

AI is predicted to use as much energy as Japan by 2026. It uses energy in two ways. First, it takes huge amounts of computer power to process the data. Training one hour of ChatGPT data uses as much energy as 80,000 households do in a year. Then, every query uses a much smaller amount of energy to produce an answer, but since there are billions of queries producing shit pictures of mutant kittens the amount of energy needed in the aggregate is vast. As I posted upthread, it also takes vast quantities of water to cool the data centres. It cannot use sea water which is too corrosive, so it takes water from the local area that disrupts the water supply for the local population, plants and animals. This is already causing social unrest in e.g. Chile. www.wired.com/story/ai-energy-demands-water-impact-internet-hyper-consumption-era/

It also requires rare metals that are mined in dangerous conditions, often by children, in places like the Democratic Republic of Congo where 6 million people have died in a war driven in part by the fight to control the rare resources. https://www.prospectmagazine.co.uk/world/foreign-correspondence/64646/in-the-east-of-the-drc-a-war-is-financed-by-blood-minerals

It also pays so-called "ghost" workers in places like Kenya under $2 an hour to train the data by spending their days tagging all the content online, including bestiality and CSA content. time.com/6247678/openai-chatgpt-kenya-workers/

So no, I am not a fan.

I remember that annoying little git bouncing around with its ridiculous happiness.
I can't remember how I got rid of it but I did eventually.
I'm really old school. Taught word processing in the 80s 😳

Ah I got lucky then - I agree, it was hugely annoying having it there!

I get why people are comparing genAI to the paperclip but IMO it's more like Photoshop or CGI. You only notice it when it's badly done.

Back in the 90s Photoshop, CGI and video post-processing were really obvious. 30 years later every image you see or film you watch is at least a little touched up and sometimes the entire background has been rreplaced and unless it's a fantastical scene you don't even realise.

The people who think they can recognise AI images or text are mistaken. They can recognise crude AI generated content. They don't know how much has slipped past them.

Objectively and morally I hate it, but I have begun to wonder if it's correcting the marketplace that social media destroyed when it comes to written content online.

The rise of blogs and socials from the late 00s and the pivot to video of the last ten years or so, both did their bit for killing off long-form, written content. The 'anyone can start a blog' era flooded the market, and when all the ad money went to click-bait and videos, publishers stopped funding the good stuff.

Now AI is doing all the aggregating, blogging etc, and the genuinely good long reads survive mostly behind paywalls. So in a way, its only the actually talented people getting paid to do it again, albeit less of them, and for less money.

As someone who used to make a decent wedge (as a side gig) writing, I'm torn.

We are doomed I think. We have become very stupid, determined to wreck our planet.

I quite liked the paperclip.

At least he was friendly.

I hate AI. I used to have a really amazing job writing web content, but now they can pretty much just type in what they want and proofread it to make tweaks. No need for me.

I had the misfortune of having to read through a lot AI text for some work I was doing. Now I see it everywhere. It’s all the same, and if I see the word ‘elevate’ now it makes me want to scream! ‘Elevate your experience’ aargh!

Because the servers and computers needed to power AI have to be kept cool and uses water to do so. Same with cryptocurrency. Huge computers.

Even just the basic, run of the mill free video generators blow my mind.

The fact that you can type in 'Two year old girl wearing yellow wellies and an orange hat, holding a white cat and waving a flag' - and then a couple of minutes later have a semi-realistic video of just that, just created from nowhere, out of thin air without any specific human involvement 🤯

The images and videos scare me a bit, I remember how bad the quality of most AI images was only a couple of years back, it’s scary how fast it has progressed, especially with videos. It feels a bit dystopian.

Yes, it's well known that AI loves particular words. I've noticed it particularly loves 'ensure/ensuring' and 'usage' as well. Basically any words that people hardly ever use normally, it particularly likes for some reason - which is why it always sounds so odd and unnatural.

And the other thing it's terrible at, obviously, is creativity. So it's basically spitting out the same garbage for everyone.

How does AI damage the environment and use water? Genuine question

My favourite use of 'frivolous' AI is the AI generated review summaries on Amazon.

Obviously not infallible because it's relying on the accuracy of the reviews themselves...but ime the actual AI summary has always been an accurate representation of the reviews themselves. Very helpful for a quick overview.

The Google ai summaries are wrong a lot of the time. They have no way of knowing if a source is reliable. And they can easily get confused. For example I asked how to remove dried pritt stick. The reply came back "to remove dried pritt stick (glue) use a solvent such as nail varnish remover. " It had obviously not realised there are different types of glue. Pritt stick is water soluble so you can just rinse it in water.

Another example today was of when a particular class of drug became available. It came back "1960s". I knew that was wrong as it was much earlier. Google had used an academic paper from a dubious journal to come up with 1960 but the right answer is the 1940s or maybe even earlier.

I use chat gpt for coding at work and it's good for the common stuff where there's loads of stuff online. It's less good for obscure things.

How? I have not started mine 🙈

It would be interesting to know what the energy consumption used to generate a throwaway image, like Jesus and his 6 fingers above vs an actual difficult question for AI is.

I can’t imagine 6 finger Jesus actually takes that much processing power.

I once spent a couple of hours on a quiet Friday afternoon (those were the days) searching "How do I kill Mr. Clipit"). A surprising number of pages appeared....

I like AI generated religious "art", which is usually hilarious. If we feed it's brain nonsense, surely AI will remain dumb?

(JC has five fingers and one thumb on his left hand)

I’d go for either of those

It is everywhere we don't need it, being useless (bad summaries on Google searches, terrible eBay descriptions, absolutely shit customer service chat bots), and not yet doing anything that useful (intelligent toilet brush that lives under the rim would be great, maybe a wheelie bin that knows when it needs to go out with which type of rubbish)