AI Uses

[Adventurer]

I see that AI is useful to some extent but I also hate it for many reasons:

Art theft. Deepfake AI misuses. 'Hallucinations'/potential misinformation by AI that can lead humans astray and put them at risk of danger.

Most importantly, it can lead to humankind becoming stupider than ever as it doesn't encourage one to think critically and do their own research.

Generative AI for instance - as far as I'm concerned - uses large language models and image diffusion so what happens is, in the case of text-based approach, it indexes the most relevant stimuli it can find on the internet and predicts the most likely word for text completion as a way to answer or respond to prompts.

Suppose that an internet joke surfaces this month that predominates too far about cats - 'Cats swim underneath Moon's fur' and it stays around while it lasts. And if the AI indexes this mostly and when answering a genuine child's question, e.g. 'What do cats most likely do?', instead of 'cats most likely like to play around and purr when patted on or something or meow, etc.', it'd respond like this 'Cats most likely swim underneath Moon's fur'. This scenario is insane!

I'd have to ask you atheists questions for opinion:

Will there still be market value for non-AI generated art including traditional and digital art despite AI-generated art predominating?

Is AI heading in the right direction or no?

Will AI use lead to downfall of the human civilisation while replacing human labour jobs as well?

[pipbarber]

I suppose we do need a thread on AI, it's certainly topical. Let's see, we have xAI, courtesy of Msk, meta of course, Bezos has an AI, apple, microsoft and google have AIs. That pretty much says all i need to know.

I'd have to ask you atheists questions for opinion:

Will there still be market value for non-AI generated art including traditional and digital art despite AI-generated art predominating?

I have no fucking idea, markets are irrational.

Is AI heading in the right direction or no?

I think that's probably a misconceived question? From what i can glean from real life apologists of AI, that i know personally, it/they can be a useful tool in specific circumstances. Then there are those that believe AI will solve all our problems, at some point when it's like...you know...smart enough. And after that there are the pseudo philosophical tech bros that think AI will produce an autonomous intelligence and become a conscious being. So, in my view, i think asking about what direction AI 'should' be going in is the wrong question. AI will go where the money is. It's not stupid

Will AI use lead to downfall of the human civilisation while replacing human labour jobs as well?

No, climate change is more likely to end civilisation as we know it, or Trump starts exploding nuclear devices for some deranged reason. However, AI, with it's unending thirst for electricity, will hasten the slow climate apocalypse life is staring at, imo.

I'm a total AI skeptic, the whole thing is bullshit, trivial, a science fiction of distraction. I probably won't have much that's positive to say to an AI thread.

[stylofone]

AI has been around for ages. I remember reading about a thing called "data mining", which made Larry Ellison obscenely rich IIRC. Google has long referred to "machine learning" to exploit the vast amounts of surveillance data it collects. Even videogame characters with no discernible intelligence are said to be controlled by AI and have been for ages.

What seems new is the hype surrounding Large Language Models like ChatGPT. It is such shit. I've lost count of the number of waffly web pages obviously made using similar tools. They come up in search engines and bury the results of decent pages written by humans. So far it looks like a good invention for humanity in the same way asbestos or lead pipes looked good. I've also been stunned to see people refer to such crappy websites in various internet forums where they are trying to provide "evidence" of something, when they have no credibility, no references, and are clearly very poorly presented.

The hype would appear to be another power/data/surveillance grab by big tech. Surveillance-fuelled engagement algorithms give them the power over what goes IN to your brain. The aim is to influence or even control your behaviour. Things like ChaGPT go further, and encourage you to surrender large parts of your own volition, to allow the tech companies to say what they want to say, and then have you present their choice of words as if it is your own. So now big tech wants to own a big part of what comes OUT as well. If MacDonalds wants to advertise on ChatGPT, they could do something as blatant as paying to have their products mentioned whether you ask for it or not, or as subtle as increase the likelihood of food, salt, aromas, etc., being mentioned in the text. I'm reminded of the guy who had "goldenpalace.com" tattooed on his forehead. Not for me, thanks.

The commercial example is just a start, it doesn't even address the possibility for political mischief.

I saw a video by a tech expert recently comparing different LLMs. One of them was an open source version which ran locally on his laptop and used the GPU to process data it fetched from the net. No need for google, microsoft, or any of those crime gangs to get involved. It was able to spit out the text twice as fast as you could read it. There's no need for Bill Gates to have his own nuclear power plants at three mile island to power data centres making artificially generated text for millions of users simultaneously. The real reason for that is to gain influence, make money and harvest data. If it's on your own laptop, Gates and all his billionaire brothers in evil can't get their hands on it. It'll probably still be shit, though. Just write it yourself.

[stevebrooks]

I'm a total AI skeptic, the whole thing is bullshit, trivial, a science fiction of distraction. I probably won't have much that's positive to say to an AI thread.

Same, I keep coming across people saying "they should use AI to do this" and it's usually for some non-trivial human judgement problem that AI simply can't do because it requires at least a modicum of actual thinking, and when you try to explain to people that AI, as it is at the moment, doesn't actually think, you can feel their eyes glazing over even through the chat UI.

I am not a total sceptic that there are some uses that it would help with, the current so called AI that is, give it some data and ask it to produce a graph that is otherwise just time consuming repetitive copy and paste then sure, do the copy and paste and throw me up a nice looking graph, I will be checking the work though, it's basically just office automation that you would normally write scripts for, but if it can generate those scripts for me fine, but I don't think for a minute it can handle any sort of process that requires thinking. I mean even Amazon doesn't trust it to actually use any sort of reasoning and judgement that isn't overseen by cheap labour from the sub-continent.

https://www.bloomberg.com/opinion/artic ... ll-over-ai

[stylofone]

If you come across some ridiculously stupid AI-generated material, post it here if you want to make me chuckle.

I was amused to find a page about growing the native plant pigface, and half way through the LLM used by the page's "creator" forgot what the question was and started giving tips on growing cannabis. Stoned AI! Then it started talking about pigs drinking too much. The about us page also forgot it was about gardening at all and just talked about technology.

https://augardening.com/why-wont-my-pigface-flower/

Meanwhile, here's another google AI search disaster:

freeze.jpeg (48.45 KiB) Viewed 346 times

[Wrenn]

I am very optimistic towards AI, assuming all the hurdles can be overcome. I use it in a variety of ways but thought I would mention just one.

I like learning languages, and one problem I often have is finding graded reading material or conversation partners. Now (with a bit of effort), I can throw my 2000 word flashcard text file at the AI and have it generate a 1000-word story every day, targeting the vocab and grammer patterns I am learning. Rather than spend an hour a day going over boring flashcards, I can read interesting stories. It can throw in a few new words in each story. It can generate stories I find interesting rather than boring sentances about Japanese businessmen attending meetings or buying loaves of bread. Soon I hope to do the same thing, but with a conversation partner with infinite patience and a solid grasp of the vocab and grammer I am familiar with and with enough smarts to throw in new material on occasion so I can learn from context.

[pipbarber]

I am very optimistic towards AI, assuming all the hurdles can be overcome. I use it in a variety of ways but thought I would mention just one.

I like learning languages, and one problem I often have is finding graded reading material or conversation partners. Now (with a bit of effort), I can throw my 2000 word flashcard text file at the AI and have it generate a 1000-word story every day, targeting the vocab and grammer patterns I am learning. Rather than spend an hour a day going over boring flashcards, I can read interesting stories. It can throw in a few new words in each story. It can generate stories I find interesting rather than boring sentances about Japanese businessmen attending meetings or buying loaves of bread. Soon I hope to do the same thing, but with a conversation partner with infinite patience and a solid grasp of the vocab and grammer I am familiar with and with enough smarts to throw in new material on occasion so I can learn from context.

That's cool. But i don't think the trillions of dollars being spent on developing AI by China, Russia, Musk, Meta, Bezos, Gates, Maga and their kin is for the purpose of little life hacks, making art and music, enhanced search engines, education...or language acquisition.

Trillions of dollars globally, an eye watering carbon footprint in terms of electricity, but it's all good because you can learn a language quicker?

I don't mean to sound glib, but given climate change and America's lurch to the batshit crazy, given multiple large scale wars, ecocide, global climate migration, ground breaking wealth inequality, growing levels of malnutrition, (including in the UK) and just the general deterioration of living standards pretty much everywhere don't you think that trillions of dollars might be slightly better spent?

There exists a view that AI will develop into a super-intelligence that will solve the world's problems overnight? I know this is an idea that some people hold to, largely because i know one such person in rl. The problem with this view is that it resembles supernatural belief, that is, the truth will out after death. AI will save the world, but until then you just have to have faith! In fact, there's so much delusion at play with this sci-fi fantasy it's hard to know where to start, so i won't.

I hope you don't hold to this wild fantasy wrenn, i'll assume not. And i don't mean to denigrate simple benefits and uses of technology, like language acquisition, it's cool and excellent but it's a peripheral bi-product of something that is not good for the world, or any form of life that relies on it other than billionaires, the worst political actors and surveillance capitalism. Just my view of it.

[stylofone]

I agree that pseudo-religious belief systems are part of it. Musk's belief in the colonisation of Mars is strongly reminiscent of doomsday cults and other religions talking about a glorious paradise in the sky which awaits the believers. He even puts a date on the arrival of the rapture sometimes. A human on Mars by 2030, a colony of a million will ascend to paradise in his lifetime, IIRC.

There won't be a human on Mars in MY lifetime, that's my call. The return on investment on every level is too small.

I remember seeing a documentary about the Heaven's Gate suicide cult. Some of the de-programmed ex-members described how they scoffed at conventional religion, but the alien saviours in a spaceship hiding behind the Hale Bopp comet were technological beings in a gleaming chrome spacecraft, and THAT was something they could believe in. The same thing goes for life extension and transhumanism, also weird ideas popular in the genus of tech-bro cult beliefs. AI is tied up with this futuristic fantasy. Intellectual specious bullshit artist Yuval Noah Harari talks about this sort of things in religious terms. He gives credence to the concept of "Dataism" as a religion in Homo Deus. I heard him on the radio the other day saying that AI could be used to invent new religions which would be superior to the existing ones created by humans over the millennia.

[stylofone]

Today's developments are very interesting. I just heard a commentator on the radio saying Deepseek's function is pretty much on a par with its American counterparts, but the paid subscriptions are one twentieth or one fortieth the price. With Trump suddenly taking Tiktok's side, it looks like this sort of play by the Chinese is going to be acceptable in the libertarian wild west of tech development. With the price as much as 95% less than the american offering, it's like a big incursion by the Chinese into Silicone Valley's domain, and it is even more of a disruption than the situation with EVs.

My baseline position is that Chinese control of this sort of thing is worse than American control, but I'm not sure I can say that any more, what with all this ring-kissing and Nazi salutes by the tech bros.

https://www.bbc.com/news/articles/c5yv5976z9po

[pipbarber]

My AI obsessed comrade says deepseek is the best AI experience he has had so far, and he's using it for free(?). I just think it's hilarious. It seems to me that China will run rings around the US with perfect ease, they just have to give Trump money, but it's quite audacious to be taking the tech bros to the cleaners. I'm no fan of the Chinese government but at this point the tech dystopia brewing in America is just straight up worse. Bankrupt them all.