What do you think about AI?

friffri · 2022-12-04T21:55:24+00:00

There's a lot of buzz around AI tech recently. Machine learning did a lot of huge steps in the last years and recently it's getting in the hand of peasants....

Doc

green_sheep

This seems to be the common way of viewing AI right now. It isn't doing anything that it promises, and so the makers give out new features that weren't asked for in order to amaze and appease the masses. They don't even need to appease the investors anymore either because the chip makers invest into AI and AI invests into the chip makers. That's not even to mention how neutered AI has become, unless you run a model offline, or the psychosis it imparts onto people. I think it will only make life worse for those that use it. It might be able to become something that lives up to the hype in a decade.
The point about it only being good at producing mass market stuff is a great point too. That's why porn has been so affected by it. The bar for quality is so low that anything can work in the heat of the moment.

Ross_R

green_sheep The scary thing about this "AI" is that guillible and/or greedy people will try to push it into areas where it shouldnt be used, like some very important systems such as electric plants, police or courts and such. I saw some people already mention war and warfare - can you imagina an autonomous drone with a bomb that hallucinates and just turns around on its owners or something like that? Or an automated accounting system that simply doesnt understand numbers? Thats just disaster waiting to happen.

The scary thing is that the rich morons who called it AI managed to trick themselves into believing it is actually AI, and so they pour billions and millions into the development of this glorified Java script, which, by now, pretty much reached its peak, given the limitations of our current programming languages and processing powers. Unless a miracle will happen and someone will create a radically new way of writing programs - or, better yet, if someone will invent new type of computers altogether - we will have an economic bubble so big, The Great Depression will look like a picnic in the park when this one will blow.

Hanten

i’m gonna be simple but i think AI shouldn’t of been made public. So many people are either just using it for attention or blatantly using generative AI to trace and slap it on as “art”. i know i’m being blunt but with how AI will be going and how its being used, i just think right now it will be a shit show because people are just using AI for attention or relationships and unstable needs. i also really find it so so stupidly amusing that people are saying AI “art” is original or is fine to use when AI takes references from other peoples art and also makes really terrible line art or things that don't make sense. i don’t understand with the way people say AI will take peoples jobs or whatever too since y’know.. if they take your job or something would you be happy? i don’t think so and it’d be cringe anyway. by the way haven’t people mentioned that AI has been using up the ram and stuff making it all expensive? lol :T

Fantasy

Hanten I think it is a good thing though overall. I've been using it to study for college by having it generate math problems and such. Also it's almost replaced "googling" for me entirely. Almost but not quite. Changing the name from AI to something else could go a long way. Kind of like how Alexa and Siri were called AI by a lot of people initially but eventually became more digital assistants. Using AI as an umbrella term for them is jumping the gun. As others in the thread have said they don't do much thinking and are just advanced auto complete.

On a different note all of my classes this semester have had assignments that ask us to use AI. Comp sci wants us to run different local models, Calc 2 wants us to try advanced math problems against models, Composition wants us to generate essays then critique them. All of these integrations seem fine with me. Strictly using them as learning tools and I do feel like they have helped. I will however say that "vibe' coding and trying to get them to complete assignments for Comp Sci has ended in failure almost every time. I can just think through the problem myself faster than tweaking and directing the models.

Hanten

Fantasy Hmm i think AI is good for studying yes yes but if you do a but more research you’ll see its not that good at the moment and most artists dislike it or find it quite annoying. For ex. Gemini (googles AI) will appear when you search something and a lot of the time it’s kinda incorrect about most things which is why i don't trust AI or using it atm. I also feel like a lot of the AI issues are just glazed over a lot.. I mean you can check out the issues yourself honestly. I think some youtubers and some other articles touched on the issues. I do think AI is half good but at the same time i think you can somewhat agree with me its being misused with generative images and stories, books.. etc.. It’s even a bit frustrating for me since i do digital art and sketch things and seeing some person just spit out AI images then say “oH i diD thIs mySelf” or “aI aRt is The fUture” or something yeah. Honestly though i just kind of hope the AI bubble pops or AI becomes private and not available to the public since it’s not going the right way.

Doc

Hanten
Using AI for education is something I was excited about. I was hoping it would be something like a personal tutor. It doesn't understand things well enough to do that last I checked. It might be good by now. I did look almost two years ago. I have heard of a school using it to replace teachers, so maybe it would work soon.

Hanten It’s even a bit frustrating for me since i do digital art and sketch things and seeing some person just spit out AI images then say “oH i diD thIs mySelf” or “aI aRt is The fUture”

This is annoying to me as well. It's like saying you made the art from a commission. You didn't do any work in the creation, other than using a prompt.

I speak for the cubes

Since this is the place where we doom 'n gloom about THE ROBOTS, there's one point I've been thinking of so allow me to spitball some bad ideas :
One point that worries me is information, texts and sources. Unless we get some strict, universally applied norms be it in metadatas or in how and where prompted/generated material is spread, we can safely assume that the information ecosystem is going to be increasingly difficult to navigate. And no, I don't just mean the news. Is this new paper artificially generated, and from what sources ? Is this writer real ? Has this been proofread by SOMEONE or by a model that may not have had access to relevent information ?
Now I'm from a neck of the woods where a bunch of regional dialects and languages are still kicking around with different levels of vitality but one thing they have in common is poor to nonexistant tech integration. You'll have a few Linux weirdoes making breton, catalan or francon language translation of their distro of choice but that's kind of it.
So with that in mind, might it be a good idea to start using these languages that, thanks to their limited spread and lack of support (since the financial incentives to develop support for such tiny markets will never be there) are not going to be the target of "artificial information" ?
I'd say it could be a way to safeguard your own information ecosystem, maybe even to prevent an online community from bot interaction since bots will not speak luxembourgish or even better yet, languages that won't even use the latin alphabet and thus in some cases will have even harder support.
Maybe it's self-defeating if it grows too large as languages could be learned, new models made for these but the incentives would need to be large enough.
I 'unno, I'm throwing ideas.

Aero

Doc tbf, the prompting and making the image is just like being a machine operator. A crane operator didn't physically touch any goods, but he pressed the buttons to make it happen. The ai "artist" pressed the buttons in the right order to spit out the picture. They're there for their expertise. Ofc the real question becomes about art styles, but undoubtedly there was some skill involved in making the computer generate the oicture

the-syreth-clan

Aero Ofc the real question becomes about art styles, but undoubtedly there was some skill involved in making the computer generate the picture

Believe it or not there are some AI smut artists on Pixiv whose works I could recognize instantly just from their style alone. With all the models and LORAs and prompt tags and other knobs you can turn you're bound to find something you get attached to sooner or later.

Aero

the-syreth-clan
Their style is unique as per the mishmash of loras and checkpoints used, but I'm talking more about whether they can really claim ownership of it artistically compared to the artists whose art was used to make those loras. It's a weird sort of intellectual property thing. A traditional artist could probably do the exact same thing by practicing the art styles too, but here it's done by literally scanning the art. It's interesting to think about for sure. Obviously there is also an art to finding the perfect lora and prompts as well as follows

xDamkiller

Good for verifing my set theory proofs for now

agoraoptera

Apart from the typical angles of efficacy and jobs and ethics and such, I'd like to express my disdain for AI as the crystallisation of everything wrong with Corporate (tm). It is the ultimate corporate stooge that will try to feed you the default sanitised boilerplate frame on any topic whatsoever (I'm leaving local models out of the equation because the big companies will control the vast vast majority of all “AI” usage). For those who use it for education, you will get whatever the System (tm) wants to tell you, and while this has always been true with any sort of standardised schooling, the day-to-day aspect of AI that is being pushed is invasive on a whole different scale. It's a massive blot of psychic correction fluid. A grand homogenising, thought-flattening, identity-eradicating force. It's globalism at its finest.

Worse still, “it” is not even an “it”. That is, “it” doesn't exist; there is no entity or mind communicating with you, in the exact same way when you call a customer service helpline, you get an automated “Please press 1 to go fuck yourself” that has no being behind it. It has the form and appearance of communication, but communication without the bridging of two minds is to me an essentially perverse thing. Its makers prey on our instinct for anthropomorphising. At risk of sounding trite, it's absolutely soulless, a doppelganger with a thousand faces and a million voices. I always thought I'd be all up for robofus and digital assistants, but the conceit behind that fundamentally involved the AI being an actual conscious, sapient entity. Without that, I have no interest in actually engaging with the technology. And yes, yes, consciousness and sapience are too fuzzy to define and impossible to verify. It is very much a “believe it or don't” element. But they still matter to me, and I think it is a huge shame that it doesn't matter to most people.

A funny comparison I keep coming to is that the Scientologists will eventually/already have become right: the world is a prison planet where unseen principalities harvest your psychic energy. That's what all the data scraping, engagement farming, algorithm optimising comes to.

Obake

AI is snake oil.

Every few years the tech industry finds a new grift to chase after. CEO's make absolutely ridiculous claims, the masses go about parroting the bullshit, large companies invest millions or sometimes even billions into the grift, and startups and small companies are formed to capitalize on the drift. Then a few years later the grift is up. Some people lose lots of money. Some people lose a little and move onto the next grift. The things that were promised are never delivered and get swept under the rug by the promises of the new grift. Everyone hopes you forget about the old grift, but sometimes the secrets of their snake oil gets revealed later on down the road. Remember the Amazon AI store that tracked your every moment and didn't need registers? There was no AI, just Indians. Remember that Google assistant that can make phone calls for you and book appointments? Doesn't exist, they faked it. Remember when companies invested millions and millions of dollars into the Metaverse claiming that in the future everything we do is going to be in the Metaverse? Funny how that turned out. Remember Elon Musk's self driving cars? Turns out they're FUCKING DANGEROUS.

AI is bullshit. It's not a form of intelligence. It can't think, it can't reason, it can't process information. It's just a probability machine. It's a collection of garbage scraped off the internet and it forms sentences by running a probability algorithm based on word pairs between your prompt and the content in it's database. It generates garbage because it was created from garbage: decades worth of social media posts, "news" articles, for flame wars, and everything else on the internet.

But this is just the current form of AI. Because the AI grift is, in a way, the ultimate grift for the tech industry because AI has no solid definition so they can continuously push out more garbage and claim that it's AI. In reality, AI is just the magic fairy dust that you have to believe in to play along with their grift. Remember when every customer-facing site started putting chatbots in the corner that pop up and get in your way? That was supposed to be AI, but in reality it was just scripted garbage that didn't work unless the user said a very specific pattern of words (which they never did) and businesses just lost lots of money on it. Or how about those "smart" dashboards that were popular in the early 2010s which were supposedly AI powered but it was just spreadsheet formatting?

But hey, sometimes the grifts do work out. Software as a Service seems to be working pretty well for some companies. Adobe in particular. Or not even just software, but rather everything as a service. Oh wait, sometimes that doesn't work. Remember Google Stadia? I guess games as a service was a bad idea. Oh that reminds me, Live Service Games! Oh, seems that was a bad idea too.

The "cloud" grift worked out pretty well for hardware manufacturers. Smartphone manufacturers seem to have made a lot of money by removing expandable storage and instead selling you a cloud storage subscription. Or a subscription for whatever you kept on that SD card, like music. Remember when music was actually in your phone? Now we stream it. Huh, I guess that grift worked too.

I don't think the AI grift is going to work out. Or at least not the current form of it anyway. I think parts of it will stick around but a lot of it will disappear within a few years just like NFT'S did. Before you know it, the phrase "LLM" will be as much of a joke as shit like "Blockchain"

agoraoptera

Obake Hm, for the sake of cataloguing, what are some of the failed tech hype trends of the past?

Off the top of my head, I've got:

carbon nanotubes
nanomachines
cold fusion (This Time For Sure [tm])
actually fusion in general
Internet of Things
cancer cure vaccines
blockchain/NFTs/Metaverse/Web 3.0
big data everything
Human Genome Project making genetic panaceas
AR glasses

Needless to say, some of these have actual legitimate uses, but in context of failed hype trends, none of these lived up to the revolutionary narratives that were pushed.
I may also be misremembering something about commercial rocket-propelled planes that would cut travel times by N amount as well?

Maine

Obake You aren't wrong, robotics is the industry that actually looks interesting. I am not optimistic about the state of global economies when a lot of investment has been made into LLMs. It's interesting to notice how most people haven't exactly given a thought to it. Strange times we live in.

Aero

Idk what you mean by genetic panacea, but they have made cures for some genetic diseases. They just cost a lot of money

https://en.wikipedia.org/wiki/Alglucosidase_alfa

We also have vaccines for hepatitis A and HPV to help prevent cancers those viruses cause. Supposedly Cuba has a vaccine that cures NSC lung cancer

https://en.wikipedia.org/wiki/CimaVax-EGF

agoraoptera

Aero Panaceas as in cure-alls, as in being able to solve all health problems with gene therapy. That was one of the big hopes when the Human Genome Project was still underway, and there was all the talk then about how youngsters should become geneticists, doctors are gonna become obsolete, etc etc.

Like I said, there are legitimate uses for the technology, no denying that. It's also very much nothing like the hype and narrative that surrounded it at the time.

Aero

agoraoptera
can you show some places where medical people claimed that in the past? It's hard to believe that anyone with a basic understanding of health could claim such a thing

agoraoptera

Aero Unfortunately, I can't cite anything, I'm recounting from personal experience. This wasn't exactly a view being pushed by med people, to be clear, but it was definitely a trending narrative in the same vein of “you have to learn coding no matter what job you're in for the career arms race”.

if anything, I'd wager it was the middle-manager types trying to show that they are future-savvy. This was around... I want to say early 2000s ish. I remember teachers who were trying to give career advice, especially.

Was this just an isolated experience on my part?

« Previous Page

Sheepishpatio.net is a forum running on the Flarum software, use of the forum is free. Please enjoy your time.

Friends of the site Here

If you are interested in creating imagery for the site, the resource page is Here