I believe AI is going to be a net negative to society for the forseeable future. AI art is a blight on artistry as a concept, and LLMs are shunting us further into search-engine-overfit post-truth world.
But also:
Reading the OOP has made me a little angry. You can see the echo chamber forming right before your eyes. Either you see things the way OOP does with no nuance, or you stop following them and are left following AI hype-bros who'll accept you instead. It's disgustingly twitter-brained. It's a bullshit purity test that only serves your comfort over actually trying to convince anyone of anything.
Consider someone who has had some small but valued usage of AI (as a reverse dictionary, for example), but generally considers things like energy usage and intellectual property rights to be serious issues we have to face for AI to truly be a net good. What does that person hear when they read this post? "That time you used ChatGPT to recall the word 'verisimilar' makes you an evil person." is what they hear. And at that moment you've cut that person off from ever actually considering your opinion ever again. Even if you're right that's not healthy.
You can also be right for the wrong reasons. You see that a lot in the anti-AI echo chambers, people who never gave a shit about IP law suddenly pretending that they care about copyright, the whole water use thing which is closer to myth than fact, or discussions on energy usage in general.
Everyone can pick up on the vibes being off with the mainstream discourse around AI, but many can't properly articulate why and they solve that cognitive dissonance with made-up or comforting bullshit.
This makes me quite uncomfortable because that's the exact same pattern of behavior we see from reactionaries, except that what weirds them out for reasons they can't or won't say explicitly isn't tech bros but immigrants and queer people.
It's not that the datacenters don't "use" water (you'll find plenty of sources confirming that), but rather that the argument stretches the concept of "water usage" well beyond the limit of meaninglessness. Water is not electricity, it can't usually be transported very far and the impact of a pumping operation is fundamentally location-dependent. Saying "X million litres of water used for Y" is usually not useful unless you're defining the local geographic context.
Pumping aquifers in a dry area and discharging the water in a field: very bad.
Pumping from and subsequently releasing water to a lake/river: mostly harmless, though sometimes in summer the additional heat pumped into the water can be harmful depending on the size of the body of water.
The real problem is that lots of areas (especially in the US) haven't updated their water rights laws since the discovery of water tables. This is hardly a new problem, and big ag remains by far the worst offender here.
Then there's the raw materials in the supply chain... and like not to downplay it but water use is not exactly at the top of the list of environmental impacts there. Concrete is hella bad on CO2 emissions, electronics use tons of precious metals that often get strip mined and processed with little to no environmental regulation, etc.
Frankly putting "datacenter pumped water out of the river then back in" in the same aggregate figure as "local lake polluted for 300 years in China by industrial byproducts" rubs me the wrong way. These are entirely different problems that do not benefit anyone from being bastardized like this. It feels the same way to me as saying "but there are children starving in Africa!" when someone throws away some food – sure throwing away food isn't great, and it's technically on-topic, but we can see how bundling these things together isn't useful, right?
The people who hate immigrants and queer people are AI's biggest defenders. It's really no wonder that people who hate life also love the machine that replaces it.
A perfect example of the just completely delusional factoids and statistics that will spontaneously form in the hater's mind. Thank you for the demonstration.
Thanks for putting a name on that! That's actually one of the few useful purposes I've found for LLMs. Sometimes you know or deduce that some thing, device, or technique must exist. The knowledge of this thing is out there, but you simply don't know the term to search for. IMO, this is actually one of the killer features of LLMs. It works well because whatever the LLM is outputting is simply and instantly verifiable. You can describe the characteristics of something to the LLM and ask it what thing has those characteristics. Then once you have a possible name, you then look that name up in a reliable source and confirm it. Sometimes the biggest hurdle to figuring something out is just learning the name of a thing. And I've found LLMs very useful as a reverse dictionary. Thanks for putting a name on it!
Source is the commercial and academic uses I've personally seen as an academic-adjacent professional that's had to deal with this sort of stuff at my job.
What was the data you saw on what volume of requests to non-llm models as they relate to utility? I can't figure out what profession have access to this kind of statistic? It would be very useful to know, thx.
I think you've misunderstood what I was saying- I don't have spreadsheets of statistics on requests for LLM AIs vs non-LLM AIs. What I have is exposure to a significant amount of various AI users, each running different kinds of AIs, and me seeing what kind of AI they're using, and for what purposes, and how well it works or doesn't.
Generally, LLM-based stuff is really only returning 'useful' results for language-based statistical analysis, which NLP handles better, faster, and vastly cheaper. For the rest, they really don't even seem to be returning useful results- I typically see a LOT of frustration.
I'm not about to give any information that could doxx myself, but the reason I see so much of this is because I'm professionally adjacent to some supercomputers. As you can imagine, those tend to be useful for AI research :P
Ah ok that's too bad. Super computers typically don't have tensor cores though, and most LLM use is presumably client use on ready trained models which desktop or mobile cpus can manage now so it will be impossible to know then
When generative AI was first taking off, I saw it as something that could empower regular people to do things that they otherwise could not afford to. The problem, as is always the case, is capitalism immediately turned into a tool of theft and abuse. The theft of training data, the power requirements, selling it for profit, competing against those whose creations were used for training without permission or attribution, the unreliability and untrustworthiness, so many ethical and technical problems.
I still don’t have a problem with using the corpus of all human knowledge for machine learning, in theory, but we’ve ended up heading in a horrible, dystopian direction that will have no good outcomes. As we hurtle toward corporate controlled AGI with no ethical or regulatory guardrails, we are racing toward a scenario where we will be slavers or extinct, and possibly both.
When generative AI was first taking off, I saw it as something that could empower regular people to do things that they otherwise could not afford to.
Except, of course, you aren't doing anything. You are no more writing, making music, or producing art than is an art director at an ad agency is. You're telling something else to make (really shitty) art on your behalf.
*Not op but still gonna reply.
Not really? The notion that someone can own (and be entitled to control) a portion of culture is absurd. It's very frustrating to see so many people take issue with AI as "theft" as if intellectual property were something that we should support and defend instead of being the actual tool for stealing artists work ("Property is theft" and all such). And obviously data centers are not built to be environmentally sustainable (not an expert, but I assume this could be done if they cared to do so). That said, using AI to do art so humans can work is the absolute peek of a stupid fucking ideas.
eh, i'll reply too. the only reason why intellectual property exists for art is because it's essentially the only way for artists to make money under this capitalist system. while i agree that a capitalist economic system is bad and that artists should be able to make a livable wage, intellectual property on art is more of a symptom of this larger problem
I just don't think that intellectual property really achieves that. It seems to me that it is a much better tool for corporate control of art and culture than for protecting small artist. Someone who is trying to pay bills with their art probably can't afford lawyers to protect that work. That said, I don't necessarily have a better solution other than just asking people to support artist directly instead of going through corporate middlemen
yeah i definitely agree, it's not the best solution, and the law is insanely biased towards the rich. hopefully one day artists will be guaranteed a livable wage for their art
Solving points 1 and 2 will also address many ethical problems people create with AI.
I believe that information should be accessible to all. My issue is not with them training in the way they did, but their monopoly on this process. (In the very same vein as Sci-Hub makes pay-walled whitepapers accessible, cutting out the profiteering publishers.)
It must be democratized and distributed, not centralized and monetized!
The way they were trained is the way they were trained.
I dont mean to say that the ethics dont matter, but you are talking as though this isnt already present tense.
The only way to go back is basically a global EMP.
What so you actually propose that is a realistic response?
This is an actual question. To this point the only advice I've seen to come from the anti-ai crowd is "dont use it. Its bad!" And that is simply not practical.
You all sound like the people who think we are actually able to get rid of guns entirely.
I'm not sure your "this is the present" argument holds much water with me. If someone stole my work and made billions off it, I'd want justice whether it was one day or one decade later.
I also don't think "this is the way it is, suck it up" is a good argument in general. Nothing would ever improve if everyone thought like that.
Also, not practical? I don't use genAI and I'm getting along just fine.
Compute warehouses the size of football fields that consume huge amounts of electricity and water absolutely can't. They can all be found extremely easily and shut down, and it would be extremely easy to prevent more from being built.
It's a weird argument to say "we could just stop doing popular things". It shows a lack of awareness. And no, explaining this doesn't mean I'm taking sides I just recognize the current reality
It's not "popular" organically, it's being forced on us by people who are invested in the technology. The chatbots are being shoved into everything because they want to make them profitable despite being money holes, not because people want it.
Your argument is invalid, the capitalists are making money. It will continue for as long as there is money to be made. Your agreement and my agreement is unnecessary.
How do we fix the problem that makes AI something that we have to deal with.
If you're arguing that people shouldn't be upset because there's no escaping it, this is an argument in favor of capitalism. Capitalism can't be escaped either.
I appreciate you taking my question a face value, you're the only one who did. Your capitalism quote worked perfectly. I was trying to use guns as my exams of shit I can't get away from.
I'm not saying that we should rage-follow but it's also unreasonable to believe it's possible to agree with every single opinion of another person let alone another community as a whole.
I tend to agree. Mass social media was a mistake. I had way better conversations and learned way more shit from random people when I was posting on a niche metal band's fan-run message board back in the 00's. Now it's all just who can post the fastest bullshit to get the most views and clicks.
Talk about AI dumbing people down, but at least it has the ability to teach you what you want to know, if you tell it to. Social media, especially with the TikTok style of content being pushed everywhere else, is just 90% pure brain rot.
Get rid of votes and worthless Internet Points and a lot of that would vanish. Of all the things to copy from Reddit and Twitter and their ilk, voting was the dumbest thing that Lemmy copied.
Yes, that's well said. I'd also take ai over social media any day.
A while ago someone launched a social media where all the people except the user are ai. I thought it was stupid when I heard of it (still do, I wouldn't use it), but people who have, have noted how different it was because "people" on it were not mainly assholes like on normal social media. The difference shows how toxic social media is.
Occasional disagreement isn’t a bad thing. Provided that the opinions expressed aren’t toxic or dangerous, what’s wrong with hearing an opinion that differs from your own? You don’t have to endorse it, share it, or even comment about it.
No two people are going to agree 100% on everything. Listening to those who disagree with you means having opportunities to learn something new, and to maybe even improve yourself based on new information.
You follow them because you're interested in their posts and you generally agree on most things. If I follow someone and they start saying FF14 is a good game im not going to unfollow just because I disagree.
This is why when learning guitar I looked up guitar lessons and then looked for people who didn't believe learning to play guitar was possible at all and the abilities instead were based upon innate talent and genetics! /s
Seriously, if learning was done by discord, then US politics (and cable news viewers) would be full of absolute scholars, instead of, you know, the exact fucking opposite of that.
after your comment, i went back to the top of this post and started reading all the comnents. It's very interesting to read the arguments from many sides and see the nuances some people bring to the conversation.
Relatedly, if you think social media threads are a great way to learn stuff I don't know what to tell you other than maybe try picking up a book and see if there's a difference there.
Coming up with a genuinely original idea is a rare skill, much harder than judging ideas is. Somebody who comes up with one good original idea (plus ninety-nine really stupid cringeworthy takes) is a better use of your reading time than somebody who reliably never gets anything too wrong, but never says anything you find new or surprising. Alyssa Vance calls this positive selection – a single good call rules you in – as opposed to negative selection, where a single bad call rules you out. You should practice positive selection for geniuses and other intellectuals.
I think about this every time I hear someone say something like “I lost all respect for Steven Pinker after he said all that stupid stuff about AI”. Your problem was thinking of “respect” as a relevant predicate to apply to Steven Pinker in the first place. Is he your father? Your youth pastor? No? Then why are you worrying about whether or not to “respect” him? Steven Pinker is a black box who occasionally spits out ideas, opinions, and arguments for you to evaluate. If some of them are arguments you wouldn’t have come up with on your own, then he’s doing you a service. If 50% of them are false, then the best-case scenario is that they’re moronically, obviously false, so that you can reject them quickly and get on with your life.
The problem mister Alexander here makes is to assume geniuses exist, or that original ideas are rare. They don't and they are not. Spend more than 15 minutes with any toddler and you'll easily reach those 100 new original ideas. Humans are new ideas machines, it's what we do. It is spontaneous, not extraneous, to us. To assume otherwise is very cynical and disingenuous. Every person has the capability to be a genius, because genius is just a social label granted to extremely narrow interpretations and projections of an individuals abilities in an extremely concrete set of skills or topic. For example, re-contextualize with a diagnosis of autism and now suddenly they are not a genius, they have an hyper-fixation.
Also, the premise that every idea, specially brand new, can be judged and ruled as good or bad in a vacuum, right out of the gate, is also very stupid. The category of genius is a very recent concoction, stemming from the halls of Victorian moral presumptions and the newly developed habit of nobility of worshiping the writings they didn't understand of people they had never met. This is what motivates the myth that genius whatever is always positive, in the popular mind. But, Goebbels was a genius at propaganda, everything that we do today in publishing is based on stuff he invented. That doesn't mean all his ideas were worth listening to, and were he alive and you followed him on Twitter (lets be honest, he would have a Twitter), that would shed a rather poor light on you.
Because, and this is the important part, humans are not a loose collection of isolated ideas. We are not modular, freely separable and reconfigurable beings. We are holistic, evolutive and integral. Sure, we might be different things to different people (privately) and audiences (publicly) at different points in time, but our own sense of identity and being is not divisible. Steven Pinker is perfectly capable of simultaneously being a liberal, atheist and intelligent linguist; a mediocre intrusionists psychologist who forgot how history works; and a stupid mysoginist and racist. All at the same time, and never stop being a single integral person. It doesn't require an imaginary score of good to bad takes ratio. That's a stupid premise. You don't keep a broken clock around in the off chance it might be right twice a day. Use a more holistic sense.
Remember, what's behind the user name is (still more often than not) a full person, not a black box (except if it is a bot, of course).
I understand and see why he didn't touched the moral aspect of his own argument. It is because any moral analysis completely dismantles his premises. Morality is the most important thing separating humans from animals and machines. Of course if someone is an evil POS it you should block and cancel their ass. It's Karl Popper all over again, if we don't rule out bad takes in the off chance there will be a good take, we end up with a Nazi bar.
Steven Pinker is a black box who occasionally spits out ideas, opinions, and arguments for you to evaluate.
This is a very weird way to look at people.
Anyone can have an original idea, not just "genuises". I don't understand outsourcing your thinking, creativity, and your right to free association because some guy had a good idea once.
(And I don't think my dad, the inventor of toasters strudle, would approve of this)
I have simpler policies. If someone I'm listening to is annoying and wrong more often than not, then I stop fucking listening to them.
I'm not sure when people started to think that they had to go about life listening to stupid opinions of annoying fuck wads they disagree with. But you absolutely do not have to live life that way.
Steven Pinker is a black box who occasionally spits out ideas, opinions, and arguments for you to evaluate. If some of them are arguments you wouldn’t have come up with on your own, then he’s doing you a service. If 50% of them are false, then the best-case scenario is that they’re moronically, obviously false, so that you can reject them quickly and get on with your life.
Yes. And. The worst-case scenario is: the black box is creating arguments deliberately designed to make you believe false things. 100% of the arguments coming out of it are false - either containing explicit falsehoods, or presenting true facts in such a way as to draw a false conclusion. If you, personally, cannot reject one of its arguments is false, it's because you lack the knowledge rhetorical skill to see how it is false.
I'm sure you can think of individuals and groups whom this applies to.
(And there's the opposite issue. An argument that is correct, but that looks incorrect to you, because your understanding of the issue is limited or incorrect already.)
The way to avoid this is to assess the trustworthiness and credibility of the black box - in other words, how much respect to give it - before assessing its arguments. Because if your black box is producing biased and manipulative arguments, assessing those arguments on their own merits, and assuming you'll be able to spot any factual inaccuracies and illogical arguments, isn't objectivity. It's arrogance.
I work at a company that uses AI to detect repirstory ilnesses in xrays and MRI scans weeks or mobths before a human doctor could.
This work has already saved thousands of peoples lives.
But good to know you anti-AI people have your 1 dimensional, 0 nuance take on the subject and are now doing moral purity tests on it and dick measuring to see who has the loudest, most extreme hatred for AI.
I mean, not really? Maybe they're both deep learning neural architectures, but one has been trained on an entire internetful of stolen creative content and the other has been trained on ethically sourced medical data. That's a pretty significant difference.
No, really. Deep learning and transformers etc. was discoveries that allowed for all of the above, just because corporate vc shitheads drag their musty balls in the latest boom abusing the piss out of it and making it uncool, does not mean the technology is a useless scam
I recently attended a congress about technology applied on healthcare.
There were works that improved diagnosis and interventions with AI, generative mainly used for synthetic data for training.
However there were also other works that left a bad aftertaste in my mouth, like replacing human interaction between the patient and a specialist with a chatbot in charge of explaining the procedure and answering questions to the patient. Some saw privacy laws as a hindrance and wanted to use any kind of private data.
Both GenAI, one that improves lives and other that improves profits.
Yeah, that's not what I was disagreeing with. You're right about that; I'm on record saying that capitalism is our first superintelligence and it's already misaligned. I'm just saying that it isn't really meaningless to object to generative AI. Sure the edges of the category are blurry, but all the LLMs and diffusion-based image generators and video generators were unethically trained on massive bodies of stolen data. Seriously, talking about AI as though the architecture is the only significant element when getting good training data is like 90% of the challenge is kind of a pet peeve of mine. And seen in that light there's a pretty significant distinction between the AI people are objecting to and the AI people aren't objecting to, and I don't think it's a matter of "a meaningless buzzword."
I totally understand that. The pet peeve of yours, i just disagree with on a fundamental level. The data is the content, and speaking about it as if the data is the technology itself is like talking about clothes in general as being useful or not. It's meaningless especially if you don't know about or acknowledge the different types of apparel and their uses. It's obviously not general knowledge but it would be like bickering about if underwear is a great idea or not, it's totally up to the individual if they want to wear them, even if being butt naked in public is illegal. If the framework is irrelevant, then the immediate problem isn't generative AI, especially the perfectly ethical open source models
Generative AI is a meaningless buzzword for the same underlying technology
What? An AI that can "detect repirstory ilnesses in xrays and MRI scans" is not generative. It does not generate anything. It's a discriminative AI. Sure, the theories behind these technologies have many things is common - but I wouldn't call them "the same underlying technology".
Because they are both computers and you can install different (GPU-bound) software on them?
It's true that generative AI is uses discriminative models behind the scenes, but the layer needed on top of that is enough to classify it as a different technology.
It's almost like it isn't the "training on a large data set" part that people hate about generative AI
ICBMs and rocket ships both burn fuel to send a payload to a destination. Why does NASA get to send tons of satellites to space, but I'm the asshole when I nuke Europe??? They both utiluze the same technology!
Nope, all generative AI is bad, no exceptions. Something that uses the same kind of technology but doesn't try to imitate a human with artistic or linguistic output isn't the kind of AI we're talking about.
All this is being stoked by OpenAI, Anthropic and such.
They want the issue to be polarized and remove any nuance, so it’s simple: use their corporate APIs, or not. Anything else is ”dangerous.”
For what they’re really scared of is awareness of locally runnable, ethical, and independent task specific tools like yours. That doesn’t make them any money. Stirring up “fuck AI” does, because that’s a battle they know they can win.
And that AI has been trained on data that has been stolen, taking away the livelihood of thousands more. Further, the environmental destruction will have the capacity to destroy millions more.
I'm not lost on the benefits; it can be used to better society. However, the lack of policy around it, especially the pandering to corporations by the American judicial system, is the crux here. For me, at least.
No. Im also part of the ethics committee at my work and since we work with peoples medical data as our training sets 9/10ths of our time is about making sure that data is collected ethically and with very specific consent.
I'm fine with that. My issue is primarily theft and permissions and the way your committee is running it should be the absolute baseline of how models gather data. Keep up the great work. I hope that this practice becomes mainstream.
Again with the conflation. They clearly mean GPTs and LLMs from the context they provide, they just don't have another name for it, mostly because people like you like to pretend that AI is shit like chatGPT when it benefits you, and regular machine learning is AI when it benefits you.
And no, GPTs are not needed, nor used, as a base for most of the useful tech, because anyone with any sense in this industry knows that good models and carefully curated training data gets you more accurate, reliable results than large amounts of shit data.
Generative AI and their outputs are derived products of their training data. I mean this ethically, not legally; I'm not a copyright lawyer.
Using the output for personal viewing (advice, science questions, or jacking off to AI porn you requested) is weird but ethical. It's equivalent to pirating a movie to watch at home.
But as soon as you show someone else the output, I consider it theft without attribution. If you generate a meme image, you're failing to attribute the artists whose work trained the AI without permission. If you generate code, that code infringes the numerous open source licenses of the training data, by failing to attribute it.
Even a simple lemmy text post generated by AI is derived from thousands of unattributed novels.
What a weird distinction. So if I get a prompt to make a particular scene in a particular artist's distinct style: not stealing. But if I share that prompt (and maybe even some seed info) to a friend, is that stealing? If I take a picture of the generated content, stealing? If someone takes it off my laptop without my knowledge are they stealing from me or the artist?
My viewpoint is that information wants to be free, and trying to restrict it is a losing battle (as shown by Ai training). The concept of IP is tenuous at best but I do recognize that artists need to eat in our capitalist reality. But once you make something and set it free to the world you inherently lose some ownership of it. Getting mad at the tech itself for the economic injustice is silly, there are plenty more important things to worry about in our hell scape.
Copyright law is more or less always formulated as limits on the rights to redistribute content, not how it is used. Hence, it isn't a particularly strange position to take that one should be allowed to do whatever one wants with gen AI in the private confines of ones home, and it is only at the moment you start to redistribute content we have to start asking the difficult questions: what is, and what is not, a derivative work of the training data? What ethical limitations, if any, should apply when we use an algorithm to effortlessly copy "a style" that another human has spent lots of effort to develop?
That makes sense wrt redistribution, but the original comment limited itself to the ethical problem and not the legal problem. I just don't see how it makes sense in that context because it's entirely unclear who owns the work, that's the nature of the technology.
If I train a model on the work of 1000 artists each of them contributes some fractional amount to each weight. When that model generates an image, it's combining a pseudorandom human token input with the weights and some random seed info.
If I provide a prompt of my own making, am I stealing 1/1000 of the content from each artist? Is the result 1/3 mine from my token input? Is the result 100% the property of whoever trained the model? Do we need to trace the traversal of the weights and sum the ownership of each artist based on their contribution to that weight? Is it nobody's due to the sheer number of random steps that convert the input intent to the final result?
No, gen AI pictures are not dirived works of their training data. They are seperate processes. The algorithm that actually generates the image has no knowledge of the training data.
The algorithms involved in the actual creation of the images are not the ones actually trained on the data. So its not at all accurate to claim they are derived.
The training data trains an algorithm that effectively just describes an image it sees (which BTW is super useful for blind people) and gives a score for each keyword.
Then the actusl generative part takes a random background, tries to denoise it into somerthing recognisable, then shows it to thr first algorithm that gives it a score on how closely it resembles the prompts. Then does some fancy maths and performs another denoising cycle and gets another score from the first algorithm, more maths, another cycle etc. Until it spits out and image that maches the prompt.
So the algorithm that genrstes the image has no data from the training process whatsoever.
But thats not the same as a derivative. Like saying a chart on which art styles were most popular in every decade is a derivate of every work in that survey. Because those works were used to create the data being presented.
I had to check to make sure I was in the right app. Rational discussion on my Lemmy? No way.
But yes. The machine can't take responsibility for shit. You hate the people and what they are doing to you. If AI didn't exist, they do it somehow else
Here. Let me make you feel more at home. Obviously all the AI data centers should be nationalized and the owners of OpenAI sent to gulags. The data centers will be requisitioned by a new state central planning committee for purposes of economic management. /s
The post is pretty clearly* about genAI, I think you're just choosing to ignore that part. There's plenty of really awesome machine learning technology that helps with disabilities, doesn't rip off artists and isn't environmentally deleterious.
The distinction between AI and GenAI is meaningless; they are buzzwords for the same underlying tech.
So is trying to bucket them based on copyright violation: there are very powerful, open dataset, more or less reproducible LLMs trained and runnable on a trivial amount of electricity you can run on your own PC right now.
Same with use cases. One can use embeddings models or tiny resnets to kill. People do, in fact, like with Palantir's generative free recognition models. At the other extreme, LLMs can be totally task focused and useless at anything else.
The distinction is corporate/enshittified vs not. Like Reddit vs Lemmy.
The distinction between AI and GenAI is meaningless; they are buzzwords for the same underlying tech.
You know this is a stupid take, right? You know that chatgpt and Stockfish, while both being forms of "artificial intelligence," are wildly incomparable, yeah? This is like saying "the distinction between an ICBM and the Saturn-V is meaningless, because they both use the same underlying tech"
That first claim makes no sense and you make no argument to back it up. The distinction is actually quite meaningful; generative AI generates new samples from an existing distribution, be it text, audio, images, or anything else. Other forms of AI solve numerous problems in different ways, such as identifying patterns we can't or inventing novel and more optimal solutions.
The distinction between AI and GenAI is like the difference between eating and cannibalism; one contains the other, but there's still a meaningful distinction.
Generative AI produces text or images by leveraging huge neural networks weighted by tons and tons of training data. It's fundamentally a system of guesses and vibes.
Machine learning in general is often much more precise. The model finding early cancer in scans isn't just guessing the next word, it's running the image through a series of precisely tuned layers.
The industry term for the distinction is supervised vs unsupervised learning.
The distinction between AI and GenAI is meaningless; they are buzzwords for the same underlying tech.
Genuinely doubt the tech used to control Zerg is the same tech used to generate an essay about elephant which contain numerous misinformation. AI lately is being used liberally, which lost their meaning.
First of all, intellectual property rights do not protect the author. I'm the author of a few papers and a book and I do not have intellectual property rights on any of these - like most of the authors I had to give them to the publishing house.
Secondly, your personal carbon footprint is bullshit.
I'm just sick of all this because we gave to "AI" too much meaning.
I don't like Generative AI tools like LLMs, image generators, voice, video etc because i see no interests in that, I think they give bad habits, and they are not understood well by their users.
Yesterday again i had to correct my mother because she told me some fun fact she had learnt by chatGPT, (that was wrong), and she refused to listen to me because "ChatGPT do plenty of researches on the net so it should know better than you".
About the thing that "it will replace artists and destroy art industry", I don't believe in that, (even if i made the choice to never use it), because it will forever be a tool. It's practical if you want a cartoony monkey image for your article (you meanie stupid journalist) but you can't say "make me a piece of art" and then put it on a museum.
Making art myself, i hate Gen AI slop from the deep of my heart but i'm obligated to admit that. (Let's not forget how it trains on copirighted media, use shitton of energy, and give no credits)
AI in others fields, like medecine, automatic subtitles, engineering, is fine for me. It won't give bad habits, it is well understood by its users, and it is truly benefical, as in being more efficient to save lifes than humans, or simply being helpful to disabled people.
TL,DR
AI in general is a tool.
Gen AI is bad as a powerful tool for everyone's use like it is bad to give to everyone an helicopter (even if it improves mobility).
AI is nonetheless a very nice tool that can save lifes and help disabled peoples IF used and understood correctly and fairly.
AI in others fields, like medecine, automatic subtitles, engineering, is fine for me. It won't give bad habits, it is well understood by its users, and it is truly benefical, as in being more efficient to save lifes than humans, or simply being helpful to disabled people.
I think the generative AI tech bros have deliberately contributed to a lot of confusion by calling all machine learning algorithms "AI".
I mean, you have some software which both works and is socially beneficial, like translation and speech recognition software.
You have some software that works, and is incredibly dangerous because it works, like facial recognition and all the horrible ways authoritarian governments can exploit it.
And then you have some software that "works" to produce socially detrimental bullshit, like generative AI.
All three of these categories use machine learning algorithms, trained on data sets to recognize and produce patterns. But they aren't the same in any other meaningful sense. Calling them all "AI" does nothing but confuse the issue.
I spent an hour talking photographs on the drive home the other night (the wife was driving and a storm have us great clouds). I was mostly playing with angles and landscape but it was fun. The kind of stuff it would take entire weeks to do thirty years ago, and I was done in an hour. I got a mediocre shot at best, but it was real dammit.
I use it to help me solve tech and code issues, but only because searching the web for help has become so bad. LLM answers are almost always better, and I hate it.
Everything is bullshit. Everything sucks. Capitalism has ruined everything.
yes, a lot of my immoral actions are because it's hard or against the grain to be more moral (e.g. being a strict vegan even when traveling or not easily accommodated, or using cars when technically I could bicycle, but on dangerous roads and long distances).
I have definitely spent most of my adult life going against the grain in extreme ways to be a "better" person, but I have been left victimized and disabled for it, so I'm trying to learn to be more moderate and not take big social problems as entirely my personal responsibility. Obviously it's not one extreme or the other, it's an interplay between personal and social / structural.
There are LLMs trained using fully open datasets that do not contain proprietary material... (CommonCorpus dataset, OLMo)
the fact that it is environmentally harmful
There are LLMs trained with minimal power (typically the same ones as above as these projects cannot afford as much resources), and local LLMs use signiciantly less power than a toaster or microwave...
the fact that it cuts back on critical, active thought
This is a usecase problem. LLMs aren't suitable for critical thinking or decision making tasks, so if it's cutting back on your "critical, active thought" you're just using it wrong anyway...
The OOP genuinely doesn't know what they're talking about and are just reacting to sensationalized rage bait on the internet lmao
Saying it uses less power that a toaster is not much. Yes, it uses less power than a thing that literally turns electricity into pure heat… but that’s sort of a requirement for toast. That’s still a LOT of electricity. And it’s not required. People don’t need to burn down a rainforest to summarize a meeting. Just use your earballs.
Saying it uses less power that a toaster is not much
Yeah but we're talking a fraction of 1%. A toaster uses 800-1500 watts for minutes, local LLM uses <300 watts for seconds. I toast something almost every day. I'd need to prompt a local LLM literally hundreds of times per day for AI to have a higher impact on the environment than my breakfast, only considering the toasting alone. I make probably around a dozen-ish prompts per week on average.
That’s still a LOT of electricity.
That's exactly my point, thanks. All kinds of appliances use loads more power than AI. We run them without thinking twice, and there's no anti-toaster movement on the internet claiming there is no ethical toast and you're an asshole for making toast without exception. If a toaster uses a ton of electricity and is acceptable, while a local LLM uses less than 1% of that, then there is no argument to be made against local LLMs on the basis of electricity use.
Your argument just doesn't hold up and could be applied to literally anything that isn't "required". Toast isn't required, you just want it. People could just stop playing video games to save more electricity, video games aren't required. People could stop using social media to save more electricity, TikTok and YouTube's servers aren't required.
People don’t need to burn down a rainforest to summarize a meeting.
I won't call your point a strawman, but you're ignoring the actual parts of LLMs that have high resource costs in order to push a narrative that doesn't reflect the full picture. These discussions need to include the initial costs to gather the dataset and most importantly for training the model.
Sure, post-training energy costs aren't worth worrying about, but I don't think people who are aware of how LLMs work were worried about that part.
It's also ignoring the absurd fucking AI datacenters that are being built with more methane turbines than they were approved for, and without any of the legally required pollution capture technology on the stacks. At least one of these datacenters is already measurably causing illness in the surrounding area.
These aren't abstract environmental damages by energy use that could potentially come from green power sources, these aren't "fraction of a toast" energy costs only caused by people running queries either.
Nope, I'm not ignoring them, but the post is specifically about exceptions. The OOP claims there are no exceptions and there is no ethical generative AI, which is false. Your comment only applies to the majority of massive LLMs hosted by massive corporations.
The CommonCorpus dataset is less than 8TB, so fits on a single hard drive, not a data center, and contains 2 trillion tokens, which is a relatively similar amount of tokens that small local LLMs are typically trained with (OLMo 2 7B and 13B were trained on 5 trilion tokens).
These local LLMs don't have high electricity use or environmental impact to train, and don't require a massive data center for training. The training cost in energy is high, but nothing like GPT4, and is only a one time cost anyway.
So, the OOP is wrong, there is ethical generative AI, trained only on data available in the public domain, and without a high environmental impact.
No, and that's irrelevant. Their post is explicitly not about the majority, but about exceptions/edge cases.
I am responding to what they posted (I even quoted them), showing that the position that "there is no ethical use for generative AI" and that there are no exceptions is provably false.
I didn't think it needed to be said because it's not relevant to this discussion, but: the majority of AI sucks on all fronts. It's bad for intellectual property, it's bad for the environment, it's bad for privacy, it's bad for people's brains, and it's bad at what it's used for.
All of these problems are not inherent to AI itself, and instead are problems with the massive short-term-profit-seeking corporations flush with unimaginable amounts of investor cash (read: unimaginable expectations and promises that they can't meet) that control the majority of AI. Once again capitalism is the real culprit, and fools like the OOP will do these strawman mental gymnastics and spread misinformation to defend capitalism at all costs.
I can get behind this clarification, so thanks for that.
I'm a realist. To that end, relevance is assigned less on the basis on pedantic deconstruction on a single post and more on the practical reality of what is unfolding around us. Are there ethical applications for generative AI? Possibly. Will they become the standard? Unlikely, given incumbent power structures that are defining and dictating long term use.
As with most things stitched into the human experience, gaming human psychology/behavioral mechanics are key to trendsetting. What the majority accepts is what reality re-acclimates to. At the moment, that appears to be mass adoption of unethical AI systems.
I don't disagree on these problems not being inherent to AI. But that sentiment has the same flavour as 'guns don't kill people' ammosexuals like to bust out when confronted.
Either way, it's clear you have a good read on what needs to happen to get all this to a better place. Hope you keep fighting to make that happen.
Yeah, agreed. But that's not what the OOP is saying in their post and their attitude and language makes me believe they're purposefully being wrong and outrageous for attention/trolling
Yeah, don't blame you for cracking the whip on hyperbole. Its good to have someone doing that to keep us sane.
What OOP is reacting to is the majority sentiment thats saturating the feed they're swimming through. It's a messy response, but the direction they're pointed in is generally correct - and a lot more aligned with your position than you might expect, despite fumbling the details.
It’s so surreal when someone posts a meme about That Guy™ doing That Thing™ and then all of a sudden That Guy™ shows up in the comments, doing That Thing™
Like, can I get your autograph? You’re famous, bro!
I sure am glad that we learned our lesson from the marketing campaigns in the 90's that pushed consumers to recycle their plastic single-use products to deflect attention away from the harm caused by their ubiquitous use in manufacturing.
Fuck those AI users for screwing over small creators and burning down the planet though. I see no problem with this framing.
I would not to get close to bike repaired by someone who is using ai to do it.
Like what the fuck xd
I am not surprised he is unable to make code work then xddd
Yes, they count, the process of making and continuing to update the underlying LLM is also what drains the lakes, they are all made on pirated info (all the big ones for sure, I've not heard of a widely available, usable model trained 100% on legally obtained data, but I suppose it could exist).
They only real exception I can think of would be to train an AI ENTIRELY on your own personally created material. No sources from other people AT ALL. Used purely for personal use, not used or available for use by the public.
Possibly, but the intention behind it is more about not exploiting other people. If it's only trained on my work, and only used by me, I'm the only one harmed by it, and that's my choice to make.
That's very deontological. Suppose you train a model that is equally good as other models, but only using your own work. (If you were a billionaire, you could commission many works to achieve this, perhaps.) Either way, you end up with an AI that allows you to produce content without hiring artists. If the end result is just as bad for artists, why is using one of those ethical?
True, but that's why I specified that it could only be used for my own personal use. Once you start publishing the output you've entered unethical territory.
I don't see the relevance of its personal use here. If it is ethical to use your own AI for personal use, why is it unethical to use an AI trained on stolen data for personal use?
I think the public domain would be fair game as well, and the fact that AI companies don't limit themselves to those works really gives away the game. An LMM that can write in the style of Shakespeare or Dickens is impressive, but people will pay for an LLM that will write their White Lotus fan fiction for them.
So I'll be honest. I use GPT to write Python scripts for my research. I'm not a coder and I don't want to be one, but I do need to model data sometimes and I find it incredibly useful that I can tell it something in English and it can write modeling scripts in Python. It's also a great way to learn some coding basics. So please tell me why this is bad and what I should do instead.
I'd say the main ethical concern at this time, regardless of harmless use cases, is the abysmal environmental impact necessary to power centralized, commercial AI models. Refer to situations like the one in Texas. A person's use of models like ChatGPT, however small, contributes to the demand for this architecture that requires incomprehensible amounts of water, while much of the world does not have enough. In classic fashion, the U.S. government is years behind on accepting what's wrong, allowing these companies to ruin communities behind a veil of hyped-up marketing about "innovation" and beating China at another dick-measuring contest.
The other concern is that ChatGPT's ability to write your Python code for data modeling is built on the hard work of programmers who will not see a cent for their contribution to the model's training. As the adage goes, "AI allows wealth to access talent, while preventing talent from accessing wealth." But since a ridiculous amount of data goes into these models, it's an amorphous ethical issue that's understandably difficult for us to contend with, because our brains struggle to comprehend so many levels of abstraction. How harmed is each individual programmer or artist? That approach ends up being meaningless, so you have to regard it more as a class-action lawsuit, where tens of thousands have been deprived as a whole.
By my measure, this AI bubble will collapse like a dying star in the next year, because the companies have no path to profitability. I hope that shifts AI development away from these environmentally-destructive practices, and eventually we'll see legislation requiring model training to be ethically sourced (Adobe is already getting ahead of the curve on this).
As for what you can do instead, people have been running local Deepseek R1 models since earlier this year, so you could follow a guide to set one up.
I'd say this is also a bit about extremism. I mean it's not wrong to be entirely against AI. I don't think I am. For example if we managed to do it ethically, I wouldn't have much of an issue with assistance systems in cars, smart home voice assistants and machine translation. I'm more opposed the more it gets towards generative AI. And because we do it the opposite of ethical in practice. I'm not necessarily opposed because of the thing itself or towards the science behind it, but because of all the bad consequences it comes with. But people like me aren't allowed a more nuanced opinion or to draw the line somewhere unless it's a perfect 0% or 100% and I feel people expect me to take some super extreme position. I still consider myself part of the anti-AI community overall, but both sides frequently misunderstand me. So I'm still subscribed to your posts and put up with the personal hate.
(Edit: Of course the take in the screenshot is stupid, though. There are a lot of compelling arguments against AI. And whether it fixes your bike or computer code isn't a matter of opinion, and it might benefit someone but that has nothing to do with justifying cost and side-effects of AI on other people.)
It's basically "I have no creative talent or skill so I'll use this because it'll teach those artists a lesson for acting so superior." It's a completely delusional disconnect with the reality of being an actually creative person in any way. Especially one when it comes to creatives trying to earn a living while must people just want their shit for free. Go harass billionaires for their shit and leave there starving artists alone.
I use LLMs in a way that reduces social anxiety from my autism, I give it details of a strange social interaction that I could not parse on my own and ask if I should worry about it, or if I should make any kind of amends or inquiries, or if I'm over thinking something and leave it alone.
I use LLMs to bounce my own ideas off of that I'm not comfortable bouncing off someone I know IRL.
I use LLM's to role play. (all kinds)
I use LLM's to find things that I can't find via conventional research methods.
And you know what, my perspective on using it for "productive/generative" usage is nuanced. I get why artists and writers are upset, however there is nothing magical about human's and their artistic abilities and in terms of material economic impacts automation of various kinds has screwed working people in the past and generally I've seen a lot less push back.
I do think that generated images and writing is pretty bland and near worthless though without a ton of human done work atm anyway. Like, sure I could generate a video of a cat dancing on a moving bus while a nuclear bomb is going off in the background or whatever wacky shit with a simple prompt but what exactly am I even going to do with that?
Highly directed AI content that includes a lot of human work tends to actually be pretty amazing IMO.
And even though all the outrage pertains to intellectual work, this technology is going to likely result in a lot of blue collar work being automated via "embodied" neural network AI's. In fact, it may be that it was needed for this kind of automation to really take off at all. Its not just white collar work. We aren't just automating slop content and corporate purposed art. The day is coming when stuff like laundry, factory/warehouse work, and kitchen work, etc. is also all being done by robots.
Sex bots are taking work away from sex workers. Turning genuine sensuality into some kind of horrible mimicry of a genuine connection.
How am I supposed to pay my bills
Do you want me to list the techbrodude technologies that were "not going anywhere" in past decades that have effectively died outside of tiny die-hard communities still living a delusion?
Remember when the Metaverse was the next great thing that wasn't going anywhere? Remember when cryptocurrency was going to wipe out banking forevermore? Remember when NFTs were going to revolutionize artists getting paid for their work? Segway or its somehow-lamer cousin "hoverboards"? Augmented Reality? 3D TVs? Theranos? Google Wave?
Hell, just go visit the Google Graveyard for a list of "hot" technologies that withered and died on the vine. (And quite a few lame technologies that shouldn't have ever even been on the vine.)
Remember all that?
But this time the techbrodudes have it right, despite there not being a viable business model; despite every AI vendor in the world burning through money faster than dumping that same cash into a forest fire. It's not going anywhere!
Every grift has two parties: the grifter and the sucker. You're not the former.
sadly. I don't have enough money to turn this shit-hose off.
Gen AI is neat, and I use it for personal processes including code, image gen, llm/chat; but it is sooooo faaaar awaaaay from being a real game changer - while all the people poised to profit off it claim it is - that it's just insane to claim it's the next wave. evidence: all the creative (photo/art/code/etc) people who are adamantly against it and have espoused reasoning.
There's another story on my feed about a 10-year-old refactoring a code base with a LLM. Go look at the comments from actual experts that take into account things like unit tests, readability, manageability, security. Humans have more context than any AI will.
LLMs are not intelligent. They are patently not. They make shit up constantly, since that is exactly what they do. Sometimes, maybe even most of the time, the shit they make up is mostly accurate... but do you want to rely on them?
When a doctor prescribes you the wrong drug, you can sue them as a recourse. When a software company has a data breach, there is often a class-action (better than nothing) as a recourse. When an AI tells you to put glue on your pizza to hold the toppings, there is no recourse, since the AI is not a legal thing and the company disclaims all liability for its output. When an AI denies your health insurance claim because of inscrutable reasons, there is no recourse.
In the first two, there is a penalty for being wrong, which is in effect an incentive to be correct -- to be accurate, to be responsible.
In the last, as an AI llm/agent/fuckingbuzzword, there is no penalty and no incentive. The AI just is as good as its input, and half the world is fucking stupid, so if we average out all the world's input, we get "barely getting by" as a result. A coding AI is at least partially trained on random stackoverflow posts asking for help. The original code there is wrong!
Sadly, it's not going anywhere. But people who rely on it will find short-term success for long-term failure. And a society relying on it is doomed. AI relies on the creative works that already exist. If we don't make any new things, AI will stagnate and die. Where will we be then?
There are places AI/LLM/Machine-Learning can be used successfully and helpfully, but they are niche. The AI bros need to be figuring out how to quickly meet a specific need instead of trying to meet all needs at the same time. Think the early 2000-s Folding at Home, how to convince republicans to wear a fucking mask during covid, why we shouldn't just eat the billionaires*.
*Hermes-3 says cannibalism is "barbaric" in most cultures, but otherwise doesn't give convincing arguments.
AI saved my pets life. You won't convince me it's 100% all bad and there's no "right" way to use it.
The way it is trained isnt intellectual theft imo.
It only becomes intellectual theft if it is used to generate something that then competes with and takes away profits from the original creators.
Thus the intellectual theft only kicks in at generation time, but the onus is still on the AI owners for not preventing it
However if I use AI to generate anything that doesn't "compete" with anyone, then "intellectual theft" doesn't matter.
For example, when I used it to assist with diagnosing a serious issue my pet was having 2 months ago that was stumping even our vet and it got the answer right, which surprised our vet when we asked them to check a very esoteric possibility (which they dubious checked and then they were shocked to find something there.
They asked us how on earth we managed to guess to check that place of all things, how could we have known. As a result we caught the issue very early when it was easy to treat and saved our pets life
It was a gallbladder infection, and her symptoms had like 20 other more likely causes individually.
But when I punched all her symptoms into GPT, everytime, it asserted if was likely the gallbladder. It had found some papers on other animals and mammals and how gallbladder infections cause that specific combo of symptoms rarely, and encouraged us to check it out.
If you think "intellectual theft" still applies here, despite it being used to save an animals life, then you are the asshole. No one "lost" profit or business to this, no one's intellectual property was infringed, and consuming the same amount of power it takes to cook 1 pizza in my oven to save my pets life is a pretty damn good trade, in my opinion.
So, yes. I think I used AI ethically there. Fight me.
Not at tremendously less of a power cost anyways. My laptop draws 35W
5 minutes of GPT is genuinely less power consumption than several hours of my laptop being actively used to do the searching manually. Laptops burn non trivial amounts of power when in use. Anyone who has held a laptop on their lap can attest to the fact they aren't exactly running cold.
Hell even a whole day of using your mobile phone is non trivial in power consumption, they also use 8~10W or so.
Using GPT for dumb shit is arguably unethical, but only in the sense that baking cookies in the oven is. You gonna go and start yelling at people for making cookies? Cooking up one batch of cookies burns WAAAY more energy than fucking around with GPT. And yet I don't see people going around bashing people for using their ovens to cook things as a hobby.
There's no good argument against what I did, by all metrics it genuinely was the ethical choice.
The power server side for 5 minutes of chatgpt, vs the power burned browsing the internet to find the info on my own (which would take hours to manually sift through)
Thats the comparison.
Even though server side power consumption to run GPT is very high, its not so high that its more than hours and hours of a laptop usage
Personally I don’t really trust the LLMs to synthesize disparate sources.
The #1 best use case for LLMs is using them as extremely powerful fuzzy searchers on very large datasets, so stuff like hunting down published papers on topics.
Dont actually use their output as the basis for reasoning, but use it to find the original articles.
For example, as a software dev, I use them often to search for the specific documentation for what I need. I then go look at the actual documentation, but the LLM is exceptionally fast at locating the document itself for me.
Basically, using them as a powerful resource to look up and find resources is key, and was why I was able to find documentation on the symptoms of my pet so fast. It would have taken me ages to find those esoteric published papers on my own, there's so much to sift through, especially when many papers cover huge amounts of info and what Im looking for is one small piece of info in that one paper.
But with an LLM I can trim down the search space instantly to a way way smaller set, and then go through that by hand. Thousands of papers turn into a couple in a matter of seconds.
Keep talking that way if that's what helps you be happy with yourself when looking in a mirror.
The truth is that degenerative AI has no useful business model, is quite literally burning up the planet to feed having no viable business model, is killing economies while it has no useful business model, and is in general dumbing down the world, all while having no prospects for ever being a viable business.
I'm glad that your story about your dog is enough for you to burn down the planet with a clear conscience.
I hope you dont use any of the other standard quality of life features day to day that consume substantially more power per day then.
There's plenty of stuff you likely take for granted every day that you use, that burn way more fossil fuels than training GPT took.
GPT did cost a lot of power, but if you put it beside other fairly standard day-to-day things people tend to take for granted, it's a drop in the bucket.
Air conditioning, both at your home if you use it, your work if you have it, stores you visit, etc etc
Public transit
Your stove
Your microwave
Your water kettle
Your heating systems everywhere you go
Your computer
Your phone
The internet
Emergency response systems
Your clothes washer and dryer
The list goes on and on. ESPECIALLY your clothes dryer, that thing uses a massive amount of power
People seriously underestimate how much power the internet uses overall. GPT's training provides a concrete, discrete, measured amount of power one specific thing used.
Whereas the internet, as a whole, over one day, uses way more power than all of GPT's training took total. The issue is "the internet" has its power consumption broadly distributed across the entire globe, in a manner that makes it basically impossible to actually measure how much "total" power you are burning just browsing the web.
But it's non trivial. Every switch between you and your destination is burning in the range of 150 watts, easily, every router is burning easily 80 watts, etc etc.
And theres dozens of those between you and 1 given destination. The process of routing your packets from your machine all the way across countries at the speed of light, and then a response back, takes a non trivial amount of power. Theres often around 8 to 15 hops between you and the destination, and every single hop tends to have multiple machines involved in that one single packet.
Its easy to handwave that enormous power consumption away because, well, you can't see it. You aren't privvy to how much power your ISP burns every day, how much power the nameservers use, etc etc.
GPT is a non trivial chunk of power... but its not THAT much compared to all the other shit going on in the web, its genuinely just a tiny drop in the bucket.
You are extremely naive if you think using GPT makes any kind of notable shift in your total carbon footprint, it doesnt even move the dial at all.
If you actually wanna pick something as a real target for reducing your carbon footprint, the two biggest contenders are:
Use public transit, or better yet, bike/walk/run to work. If you work from home, good!
For people who proclaim to care about critical active thought to think through the issue.
Lmao, how does noone see the irony of claiming to care about active thinking while boiling the issue down to an oversimplified, black and white, "all AI is bad and all uses of it are bad".
paint is poison. I don't see people making anti-print posts. Not to diss on your antiAI zealotry ( i am your asshole, you see; i love antiAI.) I just would like to see more antiPrint posts, if the environment is your concern.
When it comes to intellectual property… are you mouthing the corporations that profit from it? Is intellectual property the solution to keep creative people alive in an exploitive economy?
it's not "whataboutism". An image maker, has to take it into consideration.
marginally few artists use natural materials. If pollution is the corcern, then there will be no movies, no touring groups, no streaming, no paintings, no prints, no this… no that.
The environment is more my area of interest, so I'm going to focus on that part.
Before I looked into AI's environmental impacts, I too had thought it might be overfocusing a bit on the wrong areas, but I didn't realize how much the order of magnitudes had changed. Before the large AI models we're seeing now, data centers weren't a major source of change in energy consumption. Overall power consumption in places like the US had been mostly level for the previous 10-20 years (up until 2020). But AI is not like most past datatcenter workloads, it is constantly high power usage. Especially for model training, it's using the equipment at full utilization for almost the entire time. It's using higher energy chips and far more chips overall. Besides training, typical datacenter workloads before high AI usage weren't super high energy per request, but that isn't true of AI either. The rapid increase in the energy consumption from it is what's driving the issue
It's causing us to delay closing of fossil fuel plants. It's making previous declining datacenter energy stop declining and go the opposite direction and projected to increase datacenter energy usage go up by 165% by 2030
In Europe too, a data center-led surge in power demand is under way, after 15 years of decline in the power sector. Having surveyed utilities across the continent, Goldman Sachs Research found that the number of connection requests received by power distribution operators (a leading indicator of future demand) has risen exponentially over the past couple of years, mostly driven by data centers.
If we were talking about water usage of AI and someone brought up agriculture's (especially animal agriculture) more dominant use, that would be fair to mention and talk about. But that doesn't excuse AI's water usage, just pose another area to also focus on
I believe AI is going to be a net negative to society for the forseeable future. AI art is a blight on artistry as a concept, and LLMs are shunting us further into search-engine-overfit post-truth world.
But also:
Reading the OOP has made me a little angry. You can see the echo chamber forming right before your eyes. Either you see things the way OOP does with no nuance, or you stop following them and are left following AI hype-bros who'll accept you instead. It's disgustingly twitter-brained. It's a bullshit purity test that only serves your comfort over actually trying to convince anyone of anything.
Consider someone who has had some small but valued usage of AI (as a reverse dictionary, for example), but generally considers things like energy usage and intellectual property rights to be serious issues we have to face for AI to truly be a net good. What does that person hear when they read this post? "That time you used ChatGPT to recall the word 'verisimilar' makes you an evil person." is what they hear. And at that moment you've cut that person off from ever actually considering your opinion ever again. Even if you're right that's not healthy.
I’m a what most people would consider an AI Luddite/hater and think OOP communicates like a dogmatic asshole.
You can also be right for the wrong reasons. You see that a lot in the anti-AI echo chambers, people who never gave a shit about IP law suddenly pretending that they care about copyright, the whole water use thing which is closer to myth than fact, or discussions on energy usage in general.
Everyone can pick up on the vibes being off with the mainstream discourse around AI, but many can't properly articulate why and they solve that cognitive dissonance with made-up or comforting bullshit.
This makes me quite uncomfortable because that's the exact same pattern of behavior we see from reactionaries, except that what weirds them out for reasons they can't or won't say explicitly isn't tech bros but immigrants and queer people.
Out of curiosity, could you link a source vis-a-vis AI's water consumption?
It's not that the datacenters don't "use" water (you'll find plenty of sources confirming that), but rather that the argument stretches the concept of "water usage" well beyond the limit of meaninglessness. Water is not electricity, it can't usually be transported very far and the impact of a pumping operation is fundamentally location-dependent. Saying "X million litres of water used for Y" is usually not useful unless you're defining the local geographic context.
Pumping aquifers in a dry area and discharging the water in a field: very bad.
Pumping from and subsequently releasing water to a lake/river: mostly harmless, though sometimes in summer the additional heat pumped into the water can be harmful depending on the size of the body of water.
The real problem is that lots of areas (especially in the US) haven't updated their water rights laws since the discovery of water tables. This is hardly a new problem, and big ag remains by far the worst offender here.
Then there's the raw materials in the supply chain... and like not to downplay it but water use is not exactly at the top of the list of environmental impacts there. Concrete is hella bad on CO2 emissions, electronics use tons of precious metals that often get strip mined and processed with little to no environmental regulation, etc.
Frankly putting "datacenter pumped water out of the river then back in" in the same aggregate figure as "local lake polluted for 300 years in China by industrial byproducts" rubs me the wrong way. These are entirely different problems that do not benefit anyone from being bastardized like this. It feels the same way to me as saying "but there are children starving in Africa!" when someone throws away some food – sure throwing away food isn't great, and it's technically on-topic, but we can see how bundling these things together isn't useful, right?
The people who hate immigrants and queer people are AI's biggest defenders. It's really no wonder that people who hate life also love the machine that replaces it.
A perfect example of the just completely delusional factoids and statistics that will spontaneously form in the hater's mind. Thank you for the demonstration.
Thanks for putting a name on that! That's actually one of the few useful purposes I've found for LLMs. Sometimes you know or deduce that some thing, device, or technique must exist. The knowledge of this thing is out there, but you simply don't know the term to search for. IMO, this is actually one of the killer features of LLMs. It works well because whatever the LLM is outputting is simply and instantly verifiable. You can describe the characteristics of something to the LLM and ask it what thing has those characteristics. Then once you have a possible name, you then look that name up in a reliable source and confirm it. Sometimes the biggest hurdle to figuring something out is just learning the name of a thing. And I've found LLMs very useful as a reverse dictionary. Thanks for putting a name on it!
Using chatGPT to recall the word 'verisimilar' is an absurd waste of time, energy, and in no way justifies the use of AI.
90% of LLM/GPT use is a waste or could be done with better with another tool, including non-LLM AIs. The remaining 10% are just outright evil.
Where is your source? It sounds unbelievable
Source is the commercial and academic uses I've personally seen as an academic-adjacent professional that's had to deal with this sort of stuff at my job.
What was the data you saw on what volume of requests to non-llm models as they relate to utility? I can't figure out what profession have access to this kind of statistic? It would be very useful to know, thx.
I think you've misunderstood what I was saying- I don't have spreadsheets of statistics on requests for LLM AIs vs non-LLM AIs. What I have is exposure to a significant amount of various AI users, each running different kinds of AIs, and me seeing what kind of AI they're using, and for what purposes, and how well it works or doesn't.
Generally, LLM-based stuff is really only returning 'useful' results for language-based statistical analysis, which NLP handles better, faster, and vastly cheaper. For the rest, they really don't even seem to be returning useful results- I typically see a LOT of frustration.
I'm not about to give any information that could doxx myself, but the reason I see so much of this is because I'm professionally adjacent to some supercomputers. As you can imagine, those tend to be useful for AI research :P
Ah ok that's too bad. Super computers typically don't have tensor cores though, and most LLM use is presumably client use on ready trained models which desktop or mobile cpus can manage now so it will be impossible to know then
yyyyes they do have tensor cores? Where did you get such an absurd idea from?
My issues are fundsmentally two fold with gen AI:
Who owns and controls it (billionares and entrenched corporations)
How it is shoehorned into everything (decision making processes, human-to-human communication, my coffee machine)
I cannot wait until finally the check is due and the AI bubble pops; folding this digital snake oil sellers' house of cards.
When generative AI was first taking off, I saw it as something that could empower regular people to do things that they otherwise could not afford to. The problem, as is always the case, is capitalism immediately turned into a tool of theft and abuse. The theft of training data, the power requirements, selling it for profit, competing against those whose creations were used for training without permission or attribution, the unreliability and untrustworthiness, so many ethical and technical problems.
I still don’t have a problem with using the corpus of all human knowledge for machine learning, in theory, but we’ve ended up heading in a horrible, dystopian direction that will have no good outcomes. As we hurtle toward corporate controlled AGI with no ethical or regulatory guardrails, we are racing toward a scenario where we will be slavers or extinct, and possibly both.
Except, of course, you aren't doing anything. You are no more writing, making music, or producing art than is an art director at an ad agency is. You're telling something else to make (really shitty) art on your behalf.
yes, it's just as bad as being a director
You really take no issue with how they were all trained?
*Not op but still gonna reply. Not really? The notion that someone can own (and be entitled to control) a portion of culture is absurd. It's very frustrating to see so many people take issue with AI as "theft" as if intellectual property were something that we should support and defend instead of being the actual tool for stealing artists work ("Property is theft" and all such). And obviously data centers are not built to be environmentally sustainable (not an expert, but I assume this could be done if they cared to do so). That said, using AI to do art so humans can work is the absolute peek of a stupid fucking ideas.
eh, i'll reply too. the only reason why intellectual property exists for art is because it's essentially the only way for artists to make money under this capitalist system. while i agree that a capitalist economic system is bad and that artists should be able to make a livable wage, intellectual property on art is more of a symptom of this larger problem
I just don't think that intellectual property really achieves that. It seems to me that it is a much better tool for corporate control of art and culture than for protecting small artist. Someone who is trying to pay bills with their art probably can't afford lawyers to protect that work. That said, I don't necessarily have a better solution other than just asking people to support artist directly instead of going through corporate middlemen
yeah i definitely agree, it's not the best solution, and the law is insanely biased towards the rich. hopefully one day artists will be guaranteed a livable wage for their art
Solving points 1 and 2 will also address many ethical problems people create with AI.
I believe that information should be accessible to all. My issue is not with them training in the way they did, but their monopoly on this process. (In the very same vein as Sci-Hub makes pay-walled whitepapers accessible, cutting out the profiteering publishers.)
It must be democratized and distributed, not centralized and monetized!
The way they were trained is the way they were trained.
I dont mean to say that the ethics dont matter, but you are talking as though this isnt already present tense.
The only way to go back is basically a global EMP.
What so you actually propose that is a realistic response?
This is an actual question. To this point the only advice I've seen to come from the anti-ai crowd is "dont use it. Its bad!" And that is simply not practical.
You all sound like the people who think we are actually able to get rid of guns entirely.
I'm not sure your "this is the present" argument holds much water with me. If someone stole my work and made billions off it, I'd want justice whether it was one day or one decade later.
I also don't think "this is the way it is, suck it up" is a good argument in general. Nothing would ever improve if everyone thought like that.
Also, not practical? I don't use genAI and I'm getting along just fine.
Okay, you know those gigantic data centers that are being built that are using all our water and electricity?
Stop building them.
Seems easy.
Just like how not buying guns is easy. For the people who get it.
Guns can be concealed and smuggled.
Compute warehouses the size of football fields that consume huge amounts of electricity and water absolutely can't. They can all be found extremely easily and shut down, and it would be extremely easy to prevent more from being built.
This isn't hard.
It's a weird argument to say "we could just stop doing popular things". It shows a lack of awareness. And no, explaining this doesn't mean I'm taking sides I just recognize the current reality
The right thing isn't always popular. Something being popular is not itself a good argument for a thing to be done.
It's not "popular" organically, it's being forced on us by people who are invested in the technology. The chatbots are being shoved into everything because they want to make them profitable despite being money holes, not because people want it.
Yeah, they shouldn't be popular. Tell all your friends.
I'd argue it's not practical to use it.
Your argument is invalid, the capitalists are making money. It will continue for as long as there is money to be made. Your agreement and my agreement is unnecessary.
How do we fix the problem that makes AI something that we have to deal with.
Sabotage, public outrage, I dunno.
If you're arguing that people shouldn't be upset because there's no escaping it, this is an argument in favor of capitalism. Capitalism can't be escaped either.
I appreciate you taking my question a face value, you're the only one who did. Your capitalism quote worked perfectly. I was trying to use guns as my exams of shit I can't get away from.
I guess. I'm still anti-capitalism, though, but in the plant a fig tree for your grandchildren kind of sense.
Nah these companies don't even make money on the whole, they burn money. So your argument is invalid, and may God have mercy on your soul! 🙏
Deranged
Are people expected not to follow anyone they disagree with?
Reading other opinions? On my echo chamber platform of choice?! /s
Follow to expose yourself to different perspectives? Sure.
But it sounds like the users in question are following with the intent to reply “you’re wrong” to everything the OP puts out.
Which… I do, sadly, expect. But I wouldn’t wish for it.
Well deserved. The OOP is wrong, and it sounds like they know it and are just trolling.
I'm not saying that we should rage-follow but it's also unreasonable to believe it's possible to agree with every single opinion of another person let alone another community as a whole.
AI is whatever, but man, has social media been mind poison.
I say we burn it all down, honestly. Including this place.
I tend to agree. Mass social media was a mistake. I had way better conversations and learned way more shit from random people when I was posting on a niche metal band's fan-run message board back in the 00's. Now it's all just who can post the fastest bullshit to get the most views and clicks.
Talk about AI dumbing people down, but at least it has the ability to teach you what you want to know, if you tell it to. Social media, especially with the TikTok style of content being pushed everywhere else, is just 90% pure brain rot.
Get rid of votes and worthless Internet Points and a lot of that would vanish. Of all the things to copy from Reddit and Twitter and their ilk, voting was the dumbest thing that Lemmy copied.
Yes, that's well said. I'd also take ai over social media any day.
A while ago someone launched a social media where all the people except the user are ai. I thought it was stupid when I heard of it (still do, I wouldn't use it), but people who have, have noted how different it was because "people" on it were not mainly assholes like on normal social media. The difference shows how toxic social media is.
Occasional disagreement isn’t a bad thing. Provided that the opinions expressed aren’t toxic or dangerous, what’s wrong with hearing an opinion that differs from your own? You don’t have to endorse it, share it, or even comment about it.
No two people are going to agree 100% on everything. Listening to those who disagree with you means having opportunities to learn something new, and to maybe even improve yourself based on new information.
keeps you informed, and it shows open-mindedness
False equivalency and strawman, nice
You follow them because you're interested in their posts and you generally agree on most things. If I follow someone and they start saying FF14 is a good game im not going to unfollow just because I disagree.
why would you follow someone you agree with?
if you want to learn, you search discord.
Searching Discord is precisely the opposite of learning. You lose knowledge every second spent on Discord.
/sThey meant
Here, I was keeping it in a drawer because I thought I wouldn't need it, but obviously I did.
I couldn't take the statement itself as sarcasm because you're not wrong lol. It would have been more obvious if you glazed Discord instead I guess.
I thought my use of capitalized Discord would be subtle but noticeable that it was a joke. I guess I was too subtle.
:) i can't know, i'm not a discord user. Apparently i prefer "losing knowledge" on lemmy
This is why when learning guitar I looked up guitar lessons and then looked for people who didn't believe learning to play guitar was possible at all and the abilities instead were based upon innate talent and genetics! /s
Seriously, if learning was done by discord, then US politics (and cable news viewers) would be full of absolute scholars, instead of, you know, the exact fucking opposite of that.
guitar example does not work :/
politicians are not genuine in their discourses. Most are there for profit and they say things that even they don't believe in 🤷
Why would listening to two sides of this help you learn anything? Hearing double the lies will teach you nothing.
after your comment, i went back to the top of this post and started reading all the comnents. It's very interesting to read the arguments from many sides and see the nuances some people bring to the conversation.
That isn't all discord.
Relatedly, if you think social media threads are a great way to learn stuff I don't know what to tell you other than maybe try picking up a book and see if there's a difference there.
Rule thinkers in, not out.
The problem mister Alexander here makes is to assume geniuses exist, or that original ideas are rare. They don't and they are not. Spend more than 15 minutes with any toddler and you'll easily reach those 100 new original ideas. Humans are new ideas machines, it's what we do. It is spontaneous, not extraneous, to us. To assume otherwise is very cynical and disingenuous. Every person has the capability to be a genius, because genius is just a social label granted to extremely narrow interpretations and projections of an individuals abilities in an extremely concrete set of skills or topic. For example, re-contextualize with a diagnosis of autism and now suddenly they are not a genius, they have an hyper-fixation.
Also, the premise that every idea, specially brand new, can be judged and ruled as good or bad in a vacuum, right out of the gate, is also very stupid. The category of genius is a very recent concoction, stemming from the halls of Victorian moral presumptions and the newly developed habit of nobility of worshiping the writings they didn't understand of people they had never met. This is what motivates the myth that genius whatever is always positive, in the popular mind. But, Goebbels was a genius at propaganda, everything that we do today in publishing is based on stuff he invented. That doesn't mean all his ideas were worth listening to, and were he alive and you followed him on Twitter (lets be honest, he would have a Twitter), that would shed a rather poor light on you.
Because, and this is the important part, humans are not a loose collection of isolated ideas. We are not modular, freely separable and reconfigurable beings. We are holistic, evolutive and integral. Sure, we might be different things to different people (privately) and audiences (publicly) at different points in time, but our own sense of identity and being is not divisible. Steven Pinker is perfectly capable of simultaneously being a liberal, atheist and intelligent linguist; a mediocre intrusionists psychologist who forgot how history works; and a stupid mysoginist and racist. All at the same time, and never stop being a single integral person. It doesn't require an imaginary score of good to bad takes ratio. That's a stupid premise. You don't keep a broken clock around in the off chance it might be right twice a day. Use a more holistic sense.
Remember, what's behind the user name is (still more often than not) a full person, not a black box (except if it is a bot, of course).
I understand and see why he didn't touched the moral aspect of his own argument. It is because any moral analysis completely dismantles his premises. Morality is the most important thing separating humans from animals and machines. Of course if someone is an evil POS it you should block and cancel their ass. It's Karl Popper all over again, if we don't rule out bad takes in the off chance there will be a good take, we end up with a Nazi bar.
pinker is a very bad guy and we should not be lionizing him for any reason
This is a very weird way to look at people.
Anyone can have an original idea, not just "genuises". I don't understand outsourcing your thinking, creativity, and your right to free association because some guy had a good idea once.
(And I don't think my dad, the inventor of toasters strudle, would approve of this)
I have simpler policies. If someone I'm listening to is annoying and wrong more often than not, then I stop fucking listening to them.
I'm not sure when people started to think that they had to go about life listening to stupid opinions of annoying fuck wads they disagree with. But you absolutely do not have to live life that way.
Yes. And. The worst-case scenario is: the black box is creating arguments deliberately designed to make you believe false things. 100% of the arguments coming out of it are false - either containing explicit falsehoods, or presenting true facts in such a way as to draw a false conclusion. If you, personally, cannot reject one of its arguments is false, it's because you lack the knowledge rhetorical skill to see how it is false.
I'm sure you can think of individuals and groups whom this applies to.
(And there's the opposite issue. An argument that is correct, but that looks incorrect to you, because your understanding of the issue is limited or incorrect already.)
The way to avoid this is to assess the trustworthiness and credibility of the black box - in other words, how much respect to give it - before assessing its arguments. Because if your black box is producing biased and manipulative arguments, assessing those arguments on their own merits, and assuming you'll be able to spot any factual inaccuracies and illogical arguments, isn't objectivity. It's arrogance.
I work at a company that uses AI to detect repirstory ilnesses in xrays and MRI scans weeks or mobths before a human doctor could.
This work has already saved thousands of peoples lives.
But good to know you anti-AI people have your 1 dimensional, 0 nuance take on the subject and are now doing moral purity tests on it and dick measuring to see who has the loudest, most extreme hatred for AI.
Nobody has a problem with this, it's generative AI that's demonic
Generative AI is a meaningless buzzword for the same underlying technology, as I kinda ranted on below.
Corporate enshittification is what's demonic. When you say fuck AI, you should really mean "fuck Sam Altman"
I mean, not really? Maybe they're both deep learning neural architectures, but one has been trained on an entire internetful of stolen creative content and the other has been trained on ethically sourced medical data. That's a pretty significant difference.
No, really. Deep learning and transformers etc. was discoveries that allowed for all of the above, just because corporate vc shitheads drag their musty balls in the latest boom abusing the piss out of it and making it uncool, does not mean the technology is a useless scam
This.
I recently attended a congress about technology applied on healthcare.
There were works that improved diagnosis and interventions with AI, generative mainly used for synthetic data for training.
However there were also other works that left a bad aftertaste in my mouth, like replacing human interaction between the patient and a specialist with a chatbot in charge of explaining the procedure and answering questions to the patient. Some saw privacy laws as a hindrance and wanted to use any kind of private data.
Both GenAI, one that improves lives and other that improves profits.
Yeah, that's not what I was disagreeing with. You're right about that; I'm on record saying that capitalism is our first superintelligence and it's already misaligned. I'm just saying that it isn't really meaningless to object to generative AI. Sure the edges of the category are blurry, but all the LLMs and diffusion-based image generators and video generators were unethically trained on massive bodies of stolen data. Seriously, talking about AI as though the architecture is the only significant element when getting good training data is like 90% of the challenge is kind of a pet peeve of mine. And seen in that light there's a pretty significant distinction between the AI people are objecting to and the AI people aren't objecting to, and I don't think it's a matter of "a meaningless buzzword."
I totally understand that. The pet peeve of yours, i just disagree with on a fundamental level. The data is the content, and speaking about it as if the data is the technology itself is like talking about clothes in general as being useful or not. It's meaningless especially if you don't know about or acknowledge the different types of apparel and their uses. It's obviously not general knowledge but it would be like bickering about if underwear is a great idea or not, it's totally up to the individual if they want to wear them, even if being butt naked in public is illegal. If the framework is irrelevant, then the immediate problem isn't generative AI, especially the perfectly ethical open source models
I think DLSS/FSR/XeSS is a good example of something that is clearly ethical and also clearly generative AI. Can't really think of many others lol
What? An AI that can "detect repirstory ilnesses in xrays and MRI scans" is not generative. It does not generate anything. It's a discriminative AI. Sure, the theories behind these technologies have many things is common - but I wouldn't call them "the same underlying technology".
It is litterally the exact same technology. If i wanted to i could turn our xray product into a image generator in less than a day.
Because they are both computers and you can install different (GPU-bound) software on them?
It's true that generative AI is uses discriminative models behind the scenes, but the layer needed on top of that is enough to classify it as a different technology.
No, I mean fuck AI. You can be included in that, if you insist.
Generative AI uses the same technology. It learns when trained on a large data set.
It's almost like it isn't the "training on a large data set" part that people hate about generative AI
ICBMs and rocket ships both burn fuel to send a payload to a destination. Why does NASA get to send tons of satellites to space, but I'm the asshole when I nuke Europe??? They both utiluze the same technology!
So would you disagree with the OP about there being no exceptions?
Nope, all generative AI is bad, no exceptions. Something that uses the same kind of technology but doesn't try to imitate a human with artistic or linguistic output isn't the kind of AI we're talking about.
Except clearly some people do. This post is very specifically saying ALL AI is bad and there is no exceptions.
Generative AI isnt a well defined concept and a lot of the tech we use is indistinguishable on a technical level from "Generstive AI"
sephirAmy explicitly said generative AI
Give me an example, and watch me distinguish it from the kind of generative AI sephirAmy is talking about
Again. Geneative AI is a meaningless term.
All this is being stoked by OpenAI, Anthropic and such.
They want the issue to be polarized and remove any nuance, so it’s simple: use their corporate APIs, or not. Anything else is ”dangerous.”
For what they’re really scared of is awareness of locally runnable, ethical, and independent task specific tools like yours. That doesn’t make them any money. Stirring up “fuck AI” does, because that’s a battle they know they can win.
nobody is trashing Visual Machine Learning to assist in medical diagnostics
cool strawman though, i like his little hat
No, when you litterally say "Fuck AI, no exceptions" you are very very expliticly covering all AI in that statement.
what do you think visual machine learning applied to medical diagnostics is exactly
does it count as "ai" if i could teach an 11th grader how to build it, because it's essentially statistically filtering legos
don't lose the thread sportschampion
Well most of my colleagues have PHDs or MDs, so good luck teaching an 11th grader to do it.
They're not even people. Who knows if that story was true. They're not conscious anymore.
Frfr
And that AI has been trained on data that has been stolen, taking away the livelihood of thousands more. Further, the environmental destruction will have the capacity to destroy millions more.
I'm not lost on the benefits; it can be used to better society. However, the lack of policy around it, especially the pandering to corporations by the American judicial system, is the crux here. For me, at least.
No. Im also part of the ethics committee at my work and since we work with peoples medical data as our training sets 9/10ths of our time is about making sure that data is collected ethically and with very specific consent.
I'm fine with that. My issue is primarily theft and permissions and the way your committee is running it should be the absolute baseline of how models gather data. Keep up the great work. I hope that this practice becomes mainstream.
Those are not GPTs or LLMs. Fuck off with your bullshit trying to conflate the two.
We actually do use Generative Pre-trained Transformers as the base for a lot of our tech. So yes they are GPTs.
And even if they werent GPTs this is a post saying all AI is bad and how there is literally no exceptions to that.
Again with the conflation. They clearly mean GPTs and LLMs from the context they provide, they just don't have another name for it, mostly because people like you like to pretend that AI is shit like chatGPT when it benefits you, and regular machine learning is AI when it benefits you.
And no, GPTs are not needed, nor used, as a base for most of the useful tech, because anyone with any sense in this industry knows that good models and carefully curated training data gets you more accurate, reliable results than large amounts of shit data.
Our whole tech stack is built off of GPTs. They are just a tool, use it badly and you grt AI slop, use it well and you can save peoples lives.
As I said, anyone with sense.
Generative AI and their outputs are derived products of their training data. I mean this ethically, not legally; I'm not a copyright lawyer.
Using the output for personal viewing (advice, science questions, or jacking off to AI porn you requested) is weird but ethical. It's equivalent to pirating a movie to watch at home.
But as soon as you show someone else the output, I consider it theft without attribution. If you generate a meme image, you're failing to attribute the artists whose work trained the AI without permission. If you generate code, that code infringes the numerous open source licenses of the training data, by failing to attribute it.
Even a simple lemmy text post generated by AI is derived from thousands of unattributed novels.
What a weird distinction. So if I get a prompt to make a particular scene in a particular artist's distinct style: not stealing. But if I share that prompt (and maybe even some seed info) to a friend, is that stealing? If I take a picture of the generated content, stealing? If someone takes it off my laptop without my knowledge are they stealing from me or the artist?
My viewpoint is that information wants to be free, and trying to restrict it is a losing battle (as shown by Ai training). The concept of IP is tenuous at best but I do recognize that artists need to eat in our capitalist reality. But once you make something and set it free to the world you inherently lose some ownership of it. Getting mad at the tech itself for the economic injustice is silly, there are plenty more important things to worry about in our hell scape.
Copyright law is more or less always formulated as limits on the rights to redistribute content, not how it is used. Hence, it isn't a particularly strange position to take that one should be allowed to do whatever one wants with gen AI in the private confines of ones home, and it is only at the moment you start to redistribute content we have to start asking the difficult questions: what is, and what is not, a derivative work of the training data? What ethical limitations, if any, should apply when we use an algorithm to effortlessly copy "a style" that another human has spent lots of effort to develop?
That makes sense wrt redistribution, but the original comment limited itself to the ethical problem and not the legal problem. I just don't see how it makes sense in that context because it's entirely unclear who owns the work, that's the nature of the technology.
If I train a model on the work of 1000 artists each of them contributes some fractional amount to each weight. When that model generates an image, it's combining a pseudorandom human token input with the weights and some random seed info.
If I provide a prompt of my own making, am I stealing 1/1000 of the content from each artist? Is the result 1/3 mine from my token input? Is the result 100% the property of whoever trained the model? Do we need to trace the traversal of the weights and sum the ownership of each artist based on their contribution to that weight? Is it nobody's due to the sheer number of random steps that convert the input intent to the final result?
Robots can not have style, so infinite ethical limitations.
No, gen AI pictures are not dirived works of their training data. They are seperate processes. The algorithm that actually generates the image has no knowledge of the training data.
The numeric weights are derived from the training material it previously ate: they're not extricable.
The algorithms involved in the actual creation of the images are not the ones actually trained on the data. So its not at all accurate to claim they are derived.
Are you arguing that the training process has no effect on the output of the model? What on earth are they doing it for, then?
Not directly no.
The training data trains an algorithm that effectively just describes an image it sees (which BTW is super useful for blind people) and gives a score for each keyword.
Then the actusl generative part takes a random background, tries to denoise it into somerthing recognisable, then shows it to thr first algorithm that gives it a score on how closely it resembles the prompts. Then does some fancy maths and performs another denoising cycle and gets another score from the first algorithm, more maths, another cycle etc. Until it spits out and image that maches the prompt.
So the algorithm that genrstes the image has no data from the training process whatsoever.
It gets a, uh, score. You wrote that yourself, I don't know how you could forget.
But thats not the same as a derivative. Like saying a chart on which art styles were most popular in every decade is a derivate of every work in that survey. Because those works were used to create the data being presented.
Do y'all hate chess engines?
If yes, cool.
If no, I think you hate tech companies more than you hate AI specifically.
Yup, as always, none of these problems are inherent to AI itself, they're all problems with capitalism.
I had to check to make sure I was in the right app. Rational discussion on my Lemmy? No way.
But yes. The machine can't take responsibility for shit. You hate the people and what they are doing to you. If AI didn't exist, they do it somehow else
I do hate those people. They built an awful technology and are subjecting the world to its cancer.
It's possible to hate people doing and things, and the tools they use to do those as well.
Here. Let me make you feel more at home. Obviously all the AI data centers should be nationalized and the owners of OpenAI sent to gulags. The data centers will be requisitioned by a new state central planning committee for purposes of economic management. /s
I fail to see the problem.
Ah, lemmy.ml.
"Speak the Devil's name and he shall appear." :D
Imagine defending private property and capitalism and then thinking you aren't the devil.
None of them are inherent to capitalism either.
Facts and history disagree
What facts and history?
The post is pretty clearly* about genAI, I think you're just choosing to ignore that part. There's plenty of really awesome machine learning technology that helps with disabilities, doesn't rip off artists and isn't environmentally deleterious.
The distinction between AI and GenAI is meaningless; they are buzzwords for the same underlying tech.
So is trying to bucket them based on copyright violation: there are very powerful, open dataset, more or less reproducible LLMs trained and runnable on a trivial amount of electricity you can run on your own PC right now.
Same with use cases. One can use embeddings models or tiny resnets to kill. People do, in fact, like with Palantir's generative free recognition models. At the other extreme, LLMs can be totally task focused and useless at anything else.
The distinction is corporate/enshittified vs not. Like Reddit vs Lemmy.
You know this is a stupid take, right? You know that chatgpt and Stockfish, while both being forms of "artificial intelligence," are wildly incomparable, yeah? This is like saying "the distinction between an ICBM and the Saturn-V is meaningless, because they both use the same underlying tech"
You know that transformer and diffussion models, while both being forms of "GenAI," are wildly incomparable, yeah?
That first claim makes no sense and you make no argument to back it up. The distinction is actually quite meaningful; generative AI generates new samples from an existing distribution, be it text, audio, images, or anything else. Other forms of AI solve numerous problems in different ways, such as identifying patterns we can't or inventing novel and more optimal solutions.
The distinction between AI and GenAI is like the difference between eating and cannibalism; one contains the other, but there's still a meaningful distinction.
Generative AI produces text or images by leveraging huge neural networks weighted by tons and tons of training data. It's fundamentally a system of guesses and vibes.
Machine learning in general is often much more precise. The model finding early cancer in scans isn't just guessing the next word, it's running the image through a series of precisely tuned layers.
The industry term for the distinction is supervised vs unsupervised learning.
It's like saying hard drives are bad because of what you can store on them
Genuinely doubt the tech used to control Zerg is the same tech used to generate an essay about elephant which contain numerous misinformation. AI lately is being used liberally, which lost their meaning.
I dunno, agree to disagree.
I lose to stockfish 2. Yes.
First of all, intellectual property rights do not protect the author. I'm the author of a few papers and a book and I do not have intellectual property rights on any of these - like most of the authors I had to give them to the publishing house.
Secondly, your personal carbon footprint is bullshit.
Thirdly, everyone in the picture is an asshole.
I'm just sick of all this because we gave to "AI" too much meaning.
I don't like Generative AI tools like LLMs, image generators, voice, video etc because i see no interests in that, I think they give bad habits, and they are not understood well by their users.
Yesterday again i had to correct my mother because she told me some fun fact she had learnt by chatGPT, (that was wrong), and she refused to listen to me because "ChatGPT do plenty of researches on the net so it should know better than you".
About the thing that "it will replace artists and destroy art industry", I don't believe in that, (even if i made the choice to never use it), because it will forever be a tool. It's practical if you want a cartoony monkey image for your article (you meanie stupid journalist) but you can't say "make me a piece of art" and then put it on a museum.
Making art myself, i hate Gen AI slop from the deep of my heart but i'm obligated to admit that. (Let's not forget how it trains on copirighted media, use shitton of energy, and give no credits)
AI in others fields, like medecine, automatic subtitles, engineering, is fine for me. It won't give bad habits, it is well understood by its users, and it is truly benefical, as in being more efficient to save lifes than humans, or simply being helpful to disabled people.
TL,DR AI in general is a tool. Gen AI is bad as a powerful tool for everyone's use like it is bad to give to everyone an helicopter (even if it improves mobility). AI is nonetheless a very nice tool that can save lifes and help disabled peoples IF used and understood correctly and fairly.
I think the generative AI tech bros have deliberately contributed to a lot of confusion by calling all machine learning algorithms "AI".
I mean, you have some software which both works and is socially beneficial, like translation and speech recognition software.
You have some software that works, and is incredibly dangerous because it works, like facial recognition and all the horrible ways authoritarian governments can exploit it.
And then you have some software that "works" to produce socially detrimental bullshit, like generative AI.
All three of these categories use machine learning algorithms, trained on data sets to recognize and produce patterns. But they aren't the same in any other meaningful sense. Calling them all "AI" does nothing but confuse the issue.
I spent an hour talking photographs on the drive home the other night (the wife was driving and a storm have us great clouds). I was mostly playing with angles and landscape but it was fun. The kind of stuff it would take entire weeks to do thirty years ago, and I was done in an hour. I got a mediocre shot at best, but it was real dammit.
I do use AI (mostly like Google), but I don't think it's justified or OK, lol - I'm the problem, and I know it.
Uh huh.
I use it to help me solve tech and code issues, but only because searching the web for help has become so bad. LLM answers are almost always better, and I hate it.
Everything is bullshit. Everything sucks. Capitalism has ruined everything.
Yeah I do plenty of shit I know is a problem. Most of it just passively from living in a consumerist society.
yes, a lot of my immoral actions are because it's hard or against the grain to be more moral (e.g. being a strict vegan even when traveling or not easily accommodated, or using cars when technically I could bicycle, but on dangerous roads and long distances).
I have definitely spent most of my adult life going against the grain in extreme ways to be a "better" person, but I have been left victimized and disabled for it, so I'm trying to learn to be more moderate and not take big social problems as entirely my personal responsibility. Obviously it's not one extreme or the other, it's an interplay between personal and social / structural.
There are LLMs trained using fully open datasets that do not contain proprietary material... (CommonCorpus dataset, OLMo)
There are LLMs trained with minimal power (typically the same ones as above as these projects cannot afford as much resources), and local LLMs use signiciantly less power than a toaster or microwave...
This is a usecase problem. LLMs aren't suitable for critical thinking or decision making tasks, so if it's cutting back on your "critical, active thought" you're just using it wrong anyway...
The OOP genuinely doesn't know what they're talking about and are just reacting to sensationalized rage bait on the internet lmao
Saying it uses less power that a toaster is not much. Yes, it uses less power than a thing that literally turns electricity into pure heat… but that’s sort of a requirement for toast. That’s still a LOT of electricity. And it’s not required. People don’t need to burn down a rainforest to summarize a meeting. Just use your earballs.
Yeah man, guess show much energy it would take to draw the 4k graphics on your phone screen in 1995?
Yeah but we're talking a fraction of 1%. A toaster uses 800-1500 watts for minutes, local LLM uses <300 watts for seconds. I toast something almost every day. I'd need to prompt a local LLM literally hundreds of times per day for AI to have a higher impact on the environment than my breakfast, only considering the toasting alone. I make probably around a dozen-ish prompts per week on average.
That's exactly my point, thanks. All kinds of appliances use loads more power than AI. We run them without thinking twice, and there's no anti-toaster movement on the internet claiming there is no ethical toast and you're an asshole for making toast without exception. If a toaster uses a ton of electricity and is acceptable, while a local LLM uses less than 1% of that, then there is no argument to be made against local LLMs on the basis of electricity use.
Your argument just doesn't hold up and could be applied to literally anything that isn't "required". Toast isn't required, you just want it. People could just stop playing video games to save more electricity, video games aren't required. People could stop using social media to save more electricity, TikTok and YouTube's servers aren't required.
Strawman
That's nothing. People aren't required to eat so much meat, it even eat so much food.
I also don't like this energy argument of anti ai, when everything else in our lives already consumes so much.
Valid
You can easily use less power in other ways, too; it's not one or the other. Let's do both.
I won't call your point a strawman, but you're ignoring the actual parts of LLMs that have high resource costs in order to push a narrative that doesn't reflect the full picture. These discussions need to include the initial costs to gather the dataset and most importantly for training the model.
Sure, post-training energy costs aren't worth worrying about, but I don't think people who are aware of how LLMs work were worried about that part.
It's also ignoring the absurd fucking AI datacenters that are being built with more methane turbines than they were approved for, and without any of the legally required pollution capture technology on the stacks. At least one of these datacenters is already measurably causing illness in the surrounding area.
These aren't abstract environmental damages by energy use that could potentially come from green power sources, these aren't "fraction of a toast" energy costs only caused by people running queries either.
Nope, I'm not ignoring them, but the post is specifically about exceptions. The OOP claims there are no exceptions and there is no ethical generative AI, which is false. Your comment only applies to the majority of massive LLMs hosted by massive corporations.
The CommonCorpus dataset is less than 8TB, so fits on a single hard drive, not a data center, and contains 2 trillion tokens, which is a relatively similar amount of tokens that small local LLMs are typically trained with (OLMo 2 7B and 13B were trained on 5 trilion tokens).
These local LLMs don't have high electricity use or environmental impact to train, and don't require a massive data center for training. The training cost in energy is high, but nothing like GPT4, and is only a one time cost anyway.
So, the OOP is wrong, there is ethical generative AI, trained only on data available in the public domain, and without a high environmental impact.
You're implying the edge cases you presented are the majority being used?
No, and that's irrelevant. Their post is explicitly not about the majority, but about exceptions/edge cases.
I am responding to what they posted (I even quoted them), showing that the position that "there is no ethical use for generative AI" and that there are no exceptions is provably false.
I didn't think it needed to be said because it's not relevant to this discussion, but: the majority of AI sucks on all fronts. It's bad for intellectual property, it's bad for the environment, it's bad for privacy, it's bad for people's brains, and it's bad at what it's used for.
All of these problems are not inherent to AI itself, and instead are problems with the massive short-term-profit-seeking corporations flush with unimaginable amounts of investor cash (read: unimaginable expectations and promises that they can't meet) that control the majority of AI. Once again capitalism is the real culprit, and fools like the OOP will do these strawman mental gymnastics and spread misinformation to defend capitalism at all costs.
I can get behind this clarification, so thanks for that.
I'm a realist. To that end, relevance is assigned less on the basis on pedantic deconstruction on a single post and more on the practical reality of what is unfolding around us. Are there ethical applications for generative AI? Possibly. Will they become the standard? Unlikely, given incumbent power structures that are defining and dictating long term use.
As with most things stitched into the human experience, gaming human psychology/behavioral mechanics are key to trendsetting. What the majority accepts is what reality re-acclimates to. At the moment, that appears to be mass adoption of unethical AI systems.
I don't disagree on these problems not being inherent to AI. But that sentiment has the same flavour as 'guns don't kill people' ammosexuals like to bust out when confronted.
Either way, it's clear you have a good read on what needs to happen to get all this to a better place. Hope you keep fighting to make that happen.
Yeah, agreed. But that's not what the OOP is saying in their post and their attitude and language makes me believe they're purposefully being wrong and outrageous for attention/trolling
Yeah, don't blame you for cracking the whip on hyperbole. Its good to have someone doing that to keep us sane.
What OOP is reacting to is the majority sentiment thats saturating the feed they're swimming through. It's a messy response, but the direction they're pointed in is generally correct - and a lot more aligned with your position than you might expect, despite fumbling the details.
I used AI to scratch my balls once. I assume this counts as ethical.
AI is a marketing term. Big Tech stole ALL data. All of it. The brazen piracy is a sign they feel untouchable. We should touch them.
It’s so surreal when someone posts a meme about That Guy™ doing That Thing™ and then all of a sudden That Guy™ shows up in the comments, doing That Thing™
Like, can I get your autograph? You’re famous, bro!
In this case, there is no such thing as "That Guy™ doing That Thing™" because it's an incorrect strawman that the OOP made up to use as rage bait
It's telling that you choose to post something like this indirectly rather than addressing any of my points directly. Hmm gee wonder why
Do you also use aws/amazon because it's just sooo best in class?? Honestly I was That Guy when I first got into infra work, so I can't really judge.
Huh?
I sure am glad that we learned our lesson from the marketing campaigns in the 90's that pushed consumers to recycle their plastic single-use products to deflect attention away from the harm caused by their ubiquitous use in manufacturing.
Fuck those AI users for screwing over small creators and burning down the planet though. I see no problem with this framing.
I would not to get close to bike repaired by someone who is using ai to do it. Like what the fuck xd I am not surprised he is unable to make code work then xddd
Do people who self-host count? Like ollama? It's not like my PC is going to drain a lake.
To that person, yeah self hosting still counts.
Ethics and morality aside.
Yes, they count, the process of making and continuing to update the underlying LLM is also what drains the lakes, they are all made on pirated info (all the big ones for sure, I've not heard of a widely available, usable model trained 100% on legally obtained data, but I suppose it could exist).
I like to read the anti ai stuff, because ultimativly a lot of criticism is valid. But by god is there a lot of adolescent whining and hyperbole.
They only real exception I can think of would be to train an AI ENTIRELY on your own personally created material. No sources from other people AT ALL. Used purely for personal use, not used or available for use by the public.
This is a very IP-brained take. This is not the reason that AI is harmful.
Possibly, but the intention behind it is more about not exploiting other people. If it's only trained on my work, and only used by me, I'm the only one harmed by it, and that's my choice to make.
That's very deontological. Suppose you train a model that is equally good as other models, but only using your own work. (If you were a billionaire, you could commission many works to achieve this, perhaps.) Either way, you end up with an AI that allows you to produce content without hiring artists. If the end result is just as bad for artists, why is using one of those ethical?
True, but that's why I specified that it could only be used for my own personal use. Once you start publishing the output you've entered unethical territory.
I don't see the relevance of its personal use here. If it is ethical to use your own AI for personal use, why is it unethical to use an AI trained on stolen data for personal use?
I think the public domain would be fair game as well, and the fact that AI companies don't limit themselves to those works really gives away the game. An LMM that can write in the style of Shakespeare or Dickens is impressive, but people will pay for an LLM that will write their White Lotus fan fiction for them.
That’s what my workplace does since 1985!
So I'll be honest. I use GPT to write Python scripts for my research. I'm not a coder and I don't want to be one, but I do need to model data sometimes and I find it incredibly useful that I can tell it something in English and it can write modeling scripts in Python. It's also a great way to learn some coding basics. So please tell me why this is bad and what I should do instead.
Didn't you read the post? You're bad and should feel bad.
I'd say the main ethical concern at this time, regardless of harmless use cases, is the abysmal environmental impact necessary to power centralized, commercial AI models. Refer to situations like the one in Texas. A person's use of models like ChatGPT, however small, contributes to the demand for this architecture that requires incomprehensible amounts of water, while much of the world does not have enough. In classic fashion, the U.S. government is years behind on accepting what's wrong, allowing these companies to ruin communities behind a veil of hyped-up marketing about "innovation" and beating China at another dick-measuring contest.
The other concern is that ChatGPT's ability to write your Python code for data modeling is built on the hard work of programmers who will not see a cent for their contribution to the model's training. As the adage goes, "AI allows wealth to access talent, while preventing talent from accessing wealth." But since a ridiculous amount of data goes into these models, it's an amorphous ethical issue that's understandably difficult for us to contend with, because our brains struggle to comprehend so many levels of abstraction. How harmed is each individual programmer or artist? That approach ends up being meaningless, so you have to regard it more as a class-action lawsuit, where tens of thousands have been deprived as a whole.
By my measure, this AI bubble will collapse like a dying star in the next year, because the companies have no path to profitability. I hope that shifts AI development away from these environmentally-destructive practices, and eventually we'll see legislation requiring model training to be ethically sourced (Adobe is already getting ahead of the curve on this).
As for what you can do instead, people have been running local Deepseek R1 models since earlier this year, so you could follow a guide to set one up.
I think sometimes it is good to replace words to reevaluate a situation.
Would "I don't want to be one" be a good argument for using ai image generation?
This is extreme
I'd say this is also a bit about extremism. I mean it's not wrong to be entirely against AI. I don't think I am. For example if we managed to do it ethically, I wouldn't have much of an issue with assistance systems in cars, smart home voice assistants and machine translation. I'm more opposed the more it gets towards generative AI. And because we do it the opposite of ethical in practice. I'm not necessarily opposed because of the thing itself or towards the science behind it, but because of all the bad consequences it comes with. But people like me aren't allowed a more nuanced opinion or to draw the line somewhere unless it's a perfect 0% or 100% and I feel people expect me to take some super extreme position. I still consider myself part of the anti-AI community overall, but both sides frequently misunderstand me. So I'm still subscribed to your posts and put up with the personal hate.
(Edit: Of course the take in the screenshot is stupid, though. There are a lot of compelling arguments against AI. And whether it fixes your bike or computer code isn't a matter of opinion, and it might benefit someone but that has nothing to do with justifying cost and side-effects of AI on other people.)
It's basically "I have no creative talent or skill so I'll use this because it'll teach those artists a lesson for acting so superior." It's a completely delusional disconnect with the reality of being an actually creative person in any way. Especially one when it comes to creatives trying to earn a living while must people just want their shit for free. Go harass billionaires for their shit and leave there starving artists alone.
so you think that artists, "actually creative people", don't use genAI?
🤔
I feel this way about people who eat meat.
I have same rant with "this is the only funny AI meme' shit.
What bluesky client is that?
It is the webpage on Firefox.
Honestly I have nothing to add
I use LLMs in a way that reduces social anxiety from my autism, I give it details of a strange social interaction that I could not parse on my own and ask if I should worry about it, or if I should make any kind of amends or inquiries, or if I'm over thinking something and leave it alone.
I use LLMs to bounce my own ideas off of that I'm not comfortable bouncing off someone I know IRL.
I use LLM's to role play. (all kinds)
I use LLM's to find things that I can't find via conventional research methods.
And you know what, my perspective on using it for "productive/generative" usage is nuanced. I get why artists and writers are upset, however there is nothing magical about human's and their artistic abilities and in terms of material economic impacts automation of various kinds has screwed working people in the past and generally I've seen a lot less push back.
I do think that generated images and writing is pretty bland and near worthless though without a ton of human done work atm anyway. Like, sure I could generate a video of a cat dancing on a moving bus while a nuclear bomb is going off in the background or whatever wacky shit with a simple prompt but what exactly am I even going to do with that?
Highly directed AI content that includes a lot of human work tends to actually be pretty amazing IMO.
And even though all the outrage pertains to intellectual work, this technology is going to likely result in a lot of blue collar work being automated via "embodied" neural network AI's. In fact, it may be that it was needed for this kind of automation to really take off at all. Its not just white collar work. We aren't just automating slop content and corporate purposed art. The day is coming when stuff like laundry, factory/warehouse work, and kitchen work, etc. is also all being done by robots.
Sex bots are taking work away from sex workers. Turning genuine sensuality into some kind of horrible mimicry of a genuine connection. How am I supposed to pay my bills
Sorry but you can deny and hate all you want, it’s not going anywhere
Neither is climate change, but we should still combat it where possible.
Funny, that. Fighting against AI could be seen as fighting against climate change, considering the large carbon footprint it has.
I think it's going to contract massively. Like nft's.
Texans are facing a water shortage due to over 900 MILLION gallons of water being used to cool AI datacenters. Do you think that's sustainable?
can't we cut down on texans if they use so much water?
doesn't Texas exist because somebody invented air conditioning?
Do you want me to list the techbrodude technologies that were "not going anywhere" in past decades that have effectively died outside of tiny die-hard communities still living a delusion?
Remember when the Metaverse was the next great thing that wasn't going anywhere? Remember when cryptocurrency was going to wipe out banking forevermore? Remember when NFTs were going to revolutionize artists getting paid for their work? Segway or its somehow-lamer cousin "hoverboards"? Augmented Reality? 3D TVs? Theranos? Google Wave?
Hell, just go visit the Google Graveyard for a list of "hot" technologies that withered and died on the vine. (And quite a few lame technologies that shouldn't have ever even been on the vine.)
Remember all that?
But this time the techbrodudes have it right, despite there not being a viable business model; despite every AI vendor in the world burning through money faster than dumping that same cash into a forest fire. It's not going anywhere!
Every grift has two parties: the grifter and the sucker. You're not the former.
sadly. I don't have enough money to turn this shit-hose off.
Gen AI is neat, and I use it for personal processes including code, image gen, llm/chat; but it is sooooo faaaar awaaaay from being a real game changer - while all the people poised to profit off it claim it is - that it's just insane to claim it's the next wave. evidence: all the creative (photo/art/code/etc) people who are adamantly against it and have espoused reasoning.
There's another story on my feed about a 10-year-old refactoring a code base with a LLM. Go look at the comments from actual experts that take into account things like unit tests, readability, manageability, security. Humans have more context than any AI will.
LLMs are not intelligent. They are patently not. They make shit up constantly, since that is exactly what they do. Sometimes, maybe even most of the time, the shit they make up is mostly accurate... but do you want to rely on them?
When a doctor prescribes you the wrong drug, you can sue them as a recourse. When a software company has a data breach, there is often a class-action (better than nothing) as a recourse. When an AI tells you to put glue on your pizza to hold the toppings, there is no recourse, since the AI is not a legal thing and the company disclaims all liability for its output. When an AI denies your health insurance claim because of inscrutable reasons, there is no recourse.
In the first two, there is a penalty for being wrong, which is in effect an incentive to be correct -- to be accurate, to be responsible.
In the last, as an AI llm/agent/fuckingbuzzword, there is no penalty and no incentive. The AI just is as good as its input, and half the world is fucking stupid, so if we average out all the world's input, we get "barely getting by" as a result. A coding AI is at least partially trained on random stackoverflow posts asking for help. The original code there is wrong!
Sadly, it's not going anywhere. But people who rely on it will find short-term success for long-term failure. And a society relying on it is doomed. AI relies on the creative works that already exist. If we don't make any new things, AI will stagnate and die. Where will we be then?
There are places AI/LLM/Machine-Learning can be used successfully and helpfully, but they are niche. The AI bros need to be figuring out how to quickly meet a specific need instead of trying to meet all needs at the same time. Think the early 2000-s Folding at Home, how to convince republicans to wear a fucking mask during covid, why we shouldn't just eat the billionaires*.
*Hermes-3 says cannibalism is "barbaric" in most cultures, but otherwise doesn't give convincing arguments.
Slopvangelical LLM thumpers.
Edit: is this being misinterpreted as a pro-AI statement?
Marx would have been pro AI and anti capitalist
AI saved my pets life. You won't convince me it's 100% all bad and there's no "right" way to use it.
The way it is trained isnt intellectual theft imo.
It only becomes intellectual theft if it is used to generate something that then competes with and takes away profits from the original creators.
Thus the intellectual theft only kicks in at generation time, but the onus is still on the AI owners for not preventing it
However if I use AI to generate anything that doesn't "compete" with anyone, then "intellectual theft" doesn't matter.
For example, when I used it to assist with diagnosing a serious issue my pet was having 2 months ago that was stumping even our vet and it got the answer right, which surprised our vet when we asked them to check a very esoteric possibility (which they dubious checked and then they were shocked to find something there.
They asked us how on earth we managed to guess to check that place of all things, how could we have known. As a result we caught the issue very early when it was easy to treat and saved our pets life
It was a gallbladder infection, and her symptoms had like 20 other more likely causes individually.
But when I punched all her symptoms into GPT, everytime, it asserted if was likely the gallbladder. It had found some papers on other animals and mammals and how gallbladder infections cause that specific combo of symptoms rarely, and encouraged us to check it out.
If you think "intellectual theft" still applies here, despite it being used to save an animals life, then you are the asshole. No one "lost" profit or business to this, no one's intellectual property was infringed, and consuming the same amount of power it takes to cook 1 pizza in my oven to save my pets life is a pretty damn good trade, in my opinion.
So, yes. I think I used AI ethically there. Fight me.
Regular search could have also surfaced that information
Not at tremendously less of a power cost anyways. My laptop draws 35W
5 minutes of GPT is genuinely less power consumption than several hours of my laptop being actively used to do the searching manually. Laptops burn non trivial amounts of power when in use. Anyone who has held a laptop on their lap can attest to the fact they aren't exactly running cold.
Hell even a whole day of using your mobile phone is non trivial in power consumption, they also use 8~10W or so.
Using GPT for dumb shit is arguably unethical, but only in the sense that baking cookies in the oven is. You gonna go and start yelling at people for making cookies? Cooking up one batch of cookies burns WAAAY more energy than fucking around with GPT. And yet I don't see people going around bashing people for using their ovens to cook things as a hobby.
There's no good argument against what I did, by all metrics it genuinely was the ethical choice.
Client side power usage for conventional Internet search is about the same as chatgpt. I'm not sure why you're talking about laptop power usage.
Conventional search is less likely to lie, though.
The power server side for 5 minutes of chatgpt, vs the power burned browsing the internet to find the info on my own (which would take hours to manually sift through)
Thats the comparison.
Even though server side power consumption to run GPT is very high, its not so high that its more than hours and hours of a laptop usage
Oh, I see the point you're making.
I assumed that the information was there to be found, and a regular search would have returned it. Thus it would not have taken hours.
Personally I don't really trust the LLMs to synthesize disparate sources.
The #1 best use case for LLMs is using them as extremely powerful fuzzy searchers on very large datasets, so stuff like hunting down published papers on topics.
Dont actually use their output as the basis for reasoning, but use it to find the original articles.
For example, as a software dev, I use them often to search for the specific documentation for what I need. I then go look at the actual documentation, but the LLM is exceptionally fast at locating the document itself for me.
Basically, using them as a powerful resource to look up and find resources is key, and was why I was able to find documentation on the symptoms of my pet so fast. It would have taken me ages to find those esoteric published papers on my own, there's so much to sift through, especially when many papers cover huge amounts of info and what Im looking for is one small piece of info in that one paper.
But with an LLM I can trim down the search space instantly to a way way smaller set, and then go through that by hand. Thousands of papers turn into a couple in a matter of seconds.
Querying the LLM is not where the dangerous energy costs have ever been. It's the cost of training the model in the first place.
The training costs effectively enter a "divide by infinity" argument given enough time.
While they continue to train models at this time, eventually you hit a point where a given model can be used in perpetuity.
Costs to train go down, whereas the usability of that model stretches on to effectively infinity.
So you hit a point where you have a one time energy cost to make the model, and an infinite timescale to use it on.
Costs to train are going up exponentially. In a few years corps are going to want a return on the investment and they're going to squeeze consumers.
I can't wait for this to happen. Any day now, surely.
The singular of data is not anecdote.
For every claim people make of how "Gen AI saved my " you can find a dozen stories of people being actively harmed by Gen AI.
Stopped watch and all that jazz.
Thats irrelevant to the discussion at hand.
That's like arguing needles were a bad invention because many people use them for heroin.
People using the tool wrong to hurt themselves doesn't mean the tool is bad, it just means better regulations and education needs to be put in place.
Keep talking that way if that's what helps you be happy with yourself when looking in a mirror.
The truth is that degenerative AI has no useful business model, is quite literally burning up the planet to feed having no viable business model, is killing economies while it has no useful business model, and is in general dumbing down the world, all while having no prospects for ever being a viable business.
I'm glad that your story about your dog is enough for you to burn down the planet with a clear conscience.
Buh-bye.
I hope you dont use any of the other standard quality of life features day to day that consume substantially more power per day then.
There's plenty of stuff you likely take for granted every day that you use, that burn way more fossil fuels than training GPT took.
GPT did cost a lot of power, but if you put it beside other fairly standard day-to-day things people tend to take for granted, it's a drop in the bucket.
The list goes on and on. ESPECIALLY your clothes dryer, that thing uses a massive amount of power
People seriously underestimate how much power the internet uses overall. GPT's training provides a concrete, discrete, measured amount of power one specific thing used.
Whereas the internet, as a whole, over one day, uses way more power than all of GPT's training took total. The issue is "the internet" has its power consumption broadly distributed across the entire globe, in a manner that makes it basically impossible to actually measure how much "total" power you are burning just browsing the web.
But it's non trivial. Every switch between you and your destination is burning in the range of 150 watts, easily, every router is burning easily 80 watts, etc etc.
And theres dozens of those between you and 1 given destination. The process of routing your packets from your machine all the way across countries at the speed of light, and then a response back, takes a non trivial amount of power. Theres often around 8 to 15 hops between you and the destination, and every single hop tends to have multiple machines involved in that one single packet.
Its easy to handwave that enormous power consumption away because, well, you can't see it. You aren't privvy to how much power your ISP burns every day, how much power the nameservers use, etc etc.
GPT is a non trivial chunk of power... but its not THAT much compared to all the other shit going on in the web, its genuinely just a tiny drop in the bucket.
You are extremely naive if you think using GPT makes any kind of notable shift in your total carbon footprint, it doesnt even move the dial at all.
If you actually wanna pick something as a real target for reducing your carbon footprint, the two biggest contenders are:
Lmfao, this is the most childish take I could possibly imagine.
You cannot avoid the problems with something by sticking your head in the sand and pretending like it doesn't exist and will go away.
What do you propose?
For people who proclaim to care about critical active thought to think through the issue.
Lmao, how does noone see the irony of claiming to care about active thinking while boiling the issue down to an oversimplified, black and white, "all AI is bad and all uses of it are bad".
I have seen sooo many flat earthers say this exact same thing.
Really convenient for you that you are apparently the only person who has ever achieved deep thought.
Lol, I'm far from it, I just recognize idiocy when I see it.
paint is poison. I don't see people making anti-print posts. Not to diss on your antiAI zealotry ( i am your asshole, you see; i love antiAI.) I just would like to see more antiPrint posts, if the environment is your concern.
When it comes to intellectual property… are you mouthing the corporations that profit from it? Is intellectual property the solution to keep creative people alive in an exploitive economy?
Whataboutism is a really lazy argument.
it's not "whataboutism". An image maker, has to take it into consideration.
marginally few artists use natural materials. If pollution is the corcern, then there will be no movies, no touring groups, no streaming, no paintings, no prints, no this… no that.
Textbook whataboutism.
AI is harmful to the environment, to which you responded, NO paint is harmful.
Now you are tripling down. Cut the crap.
You what about'd printing. It is textbook Whataboutism. Sorry to trigger you with textbooks.
:) thanks for "triggering me".
i ask for coherence. If you take it as whataboutism 🤷
Harmfulness of paint doesn't discredit environmental questions around certain uses of AI.
The environment is more my area of interest, so I'm going to focus on that part.
Before I looked into AI's environmental impacts, I too had thought it might be overfocusing a bit on the wrong areas, but I didn't realize how much the order of magnitudes had changed. Before the large AI models we're seeing now, data centers weren't a major source of change in energy consumption. Overall power consumption in places like the US had been mostly level for the previous 10-20 years (up until 2020). But AI is not like most past datatcenter workloads, it is constantly high power usage. Especially for model training, it's using the equipment at full utilization for almost the entire time. It's using higher energy chips and far more chips overall. Besides training, typical datacenter workloads before high AI usage weren't super high energy per request, but that isn't true of AI either. The rapid increase in the energy consumption from it is what's driving the issue
It's causing us to delay closing of fossil fuel plants. It's making previous declining datacenter energy stop declining and go the opposite direction and projected to increase datacenter energy usage go up by 165% by 2030
https://www.goldmansachs.com/insights/articles/ai-to-drive-165-increase-in-data-center-power-demand-by-2030
If we were talking about water usage of AI and someone brought up agriculture's (especially animal agriculture) more dominant use, that would be fair to mention and talk about. But that doesn't excuse AI's water usage, just pose another area to also focus on
Sounds like a niche just waiting for you to fill it, my friend.
:)