this post was submitted on 07 Jul 2024
1246 points (95.0% liked)
Fuck AI
1374 readers
27 users here now
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
founded 8 months ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
This is a strawman argument. AI is a tool. Like any tool, it's used for negative things and positive things. Focusing on just the negative is disingenuous at best. And focusing on AI's climate impact while completely ignoring the big picture is asinine (the oil industry knew they were the primary cause of climate change more than 60 years ago).
AI has many positive use-cases yet they are completely ignored by people who lack logic and rationality.
AI is helping physicists speed up experiments into supernovae to better understand the universe.
AI is helping doctors to expedite cancer screening rates.
AI is powering robots that can do the dishes.
AI is also helping to catch illegal fishing, tackle human trafficking, and track diseases.
Yes, ai is a tool. And the person in the screenshot is criticizing a generative gpt-like and midjorney-like ai, which has a massive impact on the climate and almost no useful results.
In your examples, as I can see, they always train their own model (supernovae research, illegal fishing) or heavily customize it and use it in close conjunction with people (cancer screenings).
And so I think we talking about two different things, so I want to clarify:
ai as in neural-network algorithm that can digest massive amounts of data and give meaningful results - absolutely is useful and, I think, the more the time will pass (and more grifters move on to other fields) the more actual useful niches and cases would be solved with neural-nets.
But, ai as in we-gonna-shove-this-bot-down-your-throut gpt-like bots trained on all the data from all the internet (mostly reddit) that struggle with basic questions, hallucinate glue on pizza, generate 6-fingered hands and are close to useless in any use-case are absolutely abismal and not worth it to ruin our climate for.
Obviously by AI they mean stuff like ChatGPT. An energy intensive toy where the goal is to get it into the hands of as many paying customers as possible. And you're doing free PR for them by associating it with useful small scale research projects. I don't think most researchers will want to associate their projects with AI now that the term has been poisoned, though they might have to because many bigwigs have been sucked into the hype. The term AI has basically existed nebulously since the beginning of computing, so whether we call one thing or another AI is basically personal taste. Companies like OpenAI have successfully attached their product to the term and have created the strongest association, so ultimately if you say AI in a contemporary context a lot of people are hearing GPT-like.
Yeah, but it doesn't really help that this is a community "Fuck AI" made as "A place for all those who loathe machine-learning...". It's like saying "I loathe Dijsktra's algorithm". The term machine learning has been used since at least the 50's and it involves a lot of elegant mathematics which all essentially just try to perform optimizations of various functions in various ways. And yet, at least in places I'm exposed to, people constantly present any instance of machine learning as useless, morally wrong, theft, ineffective compared to "traditional methods" and so on, to the point where I feel uneasy telling people that I'm doing research in that area, since there's so much hate towards the entire field, not just LLMs. It might be because of them, sure, but in my experience, the popular hating of AI is not limited to ChatGPT, corporations and the like.
It is a sad thing to see. The education system especially here in the US has really failed many people. I was always super curious and excited about machine intelligence from a young age. I was born in the mid 90's. I've been dreaming about automation myself out of work so I could create and do what I love and spend more time with the people I love, and just explore and learn and grow. As a kid I noticed two things that made adults miserable:
I went to school for CS and eventually had to drop because of personal life and mental health struggles, but I'm still interested in joining the field and open source. People sometimes make me feel really sad, and misunderstood, plus discourages me from even bothering because they're so negative. I know how we got here, but it's sad it's this predictable a reaction.
By 2014-2015 I was watching a lot of machine learning videos on YouTube, playing with style GANs etc. The fact a computer could even "apply a style" and keep the image coherent was just endlessly fascinating and it made for a lot of cool photos etc. Watching AlphaGo beat a world champion in Go using Q-learning techniques and self-play was incredible. I just love it. I love future tech and I think we can have social and economic equity and much less wealth and income inequality with all this.
A lot of people don't realize labor adds a lot to the cost of what they buy and there are only so many workers. Having even today's LLMs fully implemented and realized as agents (this is very quickly coming about) things will slowly get cheaper and better, then likely more rapidly. Software development will be cheaper. Engineers, game designers and artists will bring to life incredible things that we haven't thought up yet, and will likely use these tools/entities to enhance their workflows. Yes there will be less artist grunt work, and there will be affects on jobs. It's just not going to stop anyone from doing what they love however they like to do it. It's so odd to me.
Cheers and keep your head up. If we get this right I think people will change their tune, but probably not until they see economic and quality of life improvements. Though, I'd say machine learning and machine intelligence has added a great deal of value and opportunity in my life. I wish everyone a good future regardless of how your feel about this. I just hope people who aren't in the field or weren't enthusiastic before will at least remember there are a lot of real, kind, and extremely intelligent people working really hard on this stuff, and truly believe they can make a substantial positive impact. We should be more trusting and open. It's really hard to do it, we can get burned, but most people want decent things for most others despite disagreement and don't strife. We've made it this far. Let's go to the stars 🤠
joins community called "fuck AI"
gets mad that people there dislike AI
Let's instead make an honest attempt to de-poison the term, rather rhat just giving in. It is indeed like saying "All math bad" because math can be used in bad ways.
but those are the cool interesting research related AIs, not the venture capital hype LLMs that will gift us AGI any day now with just a bit more training data/compute.
I thought the statement was more about how much energy it costs to run AI while the planet burns.
But the reason the planet burns is because of how we generate the energy, not because of using energy. I'm not defending all these fucked up greedy corporations and their use of AI, machine learning, LLMs or whatever crap they are trying to get us to use want or not, but our real problem is based on energy generation, not consumption.
yeah it's all the evil power companies' fault.
I take it these ai's are coming up with the solution to the cheap, clean energy problem that has escaped organic intelligences for the past 50 years.
I think the EV fanatics all learned about the magic electricity grid from the same people, but i still don't see all EVs being supplied with the required photo voltaic system and secondary battery to make them load and shape neutral.
It's not my fault that all my unnecessary short haul flights contribute of global warming , the airline should have invented a clean plane.
Yeah you're absolutely right but the fact is, we're burning through energy at a massive rate to power processing data for AI.
immediately goes to whataboutism and chooses big oil as an example. Pure gold.
If you're complaining about climate impact, looking at the big picture isn't whataboutism. It's the biggest part of the dataset which is important for anyone who actually cares about the issue.
You're complaining about a single cow fart, then when someone points out there are thousands of cars with greater emissions - you cry whataboutism. It's clear you just want to have a knee-jerk reaction without rationally looking at the problem.
But these are other applications of AI. I think he meant LLMS. That would be like saying "fitting functions has many other applications and can be used for good".
If all fossil fuel power plants were converted to nuclear then tech power consumption wouldn't even matter. Again, it was the oil industry that railroaded nuclear power as being unsafe.
If we had infinite money plus infinite people with the required skills to design and build nuclear power plants plus a magical method to build nuclear reactors in 2 months (or even instantly !) plus managed to convince the public opinion that nuclear energy is actually fine, then the climate crisis would be only partially solved ! Hurray ! (This doesn't in and of itself solve food production & consumption, transportation and other sources of land use change emissions, we'd need a whole lot more work or on many other subjects)
In more serious terms (Net Zero research), nuclear isn't perfect nor is it the be all and end all solution, but it IS globally a part of the solution to generate cleaner electricity and cutting emissions. However, since we don't have all the magical things I was listing earlier, its development encounters many roadblocks and it turns out that wind and solar are extremely well scalable, integrates pretty well into grids as long as we're willing to develop the (mostly known) solutions to counter their variability (several exemples of high integration rates in different settings). The issue is that all of this (both nuclear and renewables) demands a lot of investment in terms of money, of people with the required skill sets and educating the public opinion that this is needed and desirable. And that's a MASSIVE challenge.
Which is why, to get to the point, the enormous electricity use of AI is actually a problem because its additional power consumption is keeping fossil power plants running or making them run more when emissions should be declining due to advancements in low carbon electricity production (mostly renewables). In general, it makes reaching Net Zero goals harder.
Do you have a proposed solution for nuclear waste?
We can (but largely don’t) recycle nuclear waste, completely negating the need for ultra-long term (i.e., measured in the thousands of years) storage and getting more overall energy relative to the waste that will end up in long term (measured in hundreds of years) storage.
That said, my understanding is that we have a plan for dealing with the waste, but it’s been awaiting formal review for a decade. This plan was already approved in 2002 but was shut down in 2010 for political reasons, not because of technical or safety concerns.
I do. The same thing we have been doing with it this entire time storing it in underground bunkers. Unlike the propaganda, the waste from nuclear is rather small and unlike pollution from fossil fuel plants, it is easily contained however much longer living. The benefit still out weigh the cost of managing disposal. The reality is that there is plenty of uninhabited land on the planet where nuclear waste can be stored and isolated for thousands of years. One day, hopefully, we will have fusion power which won't generate waste. And perhaps, someday we will also figure out how to permanently disposal of the nuclear waste. In the mean time, storage is a fine solution that far out weighs polluting the atmosphere with burning things.
The one we already use and works fine: cement it and bury it
Most of the people on this website hate AI without even understanding it, and refuse to make an honest assessment of its capabilities, instead pretending that it's nothing more than a good auto correct predictor engine.
Everyone makes their own risk analysis, a lot of people think that whatever you say it can do, its not worth the cost overall.
Unfortunately its your problem to disentangle useful AI from predatory AI. It would probably make sense to just call it something else (neural network, a new programming language, a new data analysis model), but then how would you trick investors?
bullshit take. OP didn't post a screenshot about AI, it's about LLMs. They are absolutely doing more harm than good. And the examples you are quoting are also highly misleading at best:
You seem to be arguing against another stawman. OP didn't say they only dislike LLM the sub is even "Fuck AI". And this thread is talking about AI in general.
Machine Learning is a subset of AI and has always been. Also, LLM is a subset of Machine Learning. You are trying to split hairs, or at least do a "That's not a real Scotsman" out of the above post.
My bad bad for not seeing the sub's name before commenting. My points still stand though.
There's machine learning, and there's machine learning. Either way, pattern matching & statistics has nothing to do with intelligence beyond the actual pattern matching logic itself. Only morons call LLMs "AI". A simple rule "if value > threshold then doSomething" is more AI than an LLM. Because there's actual logic there. An LLM has no such logic behind word prediction, but thanks to statistic it is able to fool many people (including myself, depending on the context) into believing it is intelligent. So that makes it dangerous, but not AI.
ML didn't aggressively claim the name AI as a buzzword to scam massive investment in trash. Someone talking about ML calls it ML.
Someone talking about "AI" is almost certainly not referring to ML.
Companies hve always simplified smart things and called it AI. AI is hotter than ever now, not only LLM.
And again ML is a subset of AI, LLM is a subset of ML. With these definitions, everything is AI. Look up the definition of AI. It's just a collection of techniques to do "smarter" things with AI. It includes all of the above, e.g. "If this then that" but also more advanced mathematics, like statistical methods and ML. LLM is one of those statistical models.
It doesn't matter how similar the underlying math is. LLMs and ML are wildly different in every way that matters.
ML is taking a specific data set, in one specific problem space, to model a specific problem in that one specific space. It is inherently a limited application, because that's what the math can do. It finds patterns better than our brains. It doesn't reason. ML works.
LLMs are taking a broad data set, that's primarily junk, and trying to solve far more complicated problems, generally, without any tools to do so. LLMs do not work. They confabulate.
ML has been used heavily for a long time (because it's not junk) and companies have never made a point of calling it AI. This AI bubble is all about the dumpster fire that is LLMs being wildly overused. Companies selling "AI" to investors aren't doing tried and true ML.
Yea, this bubble is mostly LLM, but also deepfakes and other generative image algorithms. They are all ML. LLM has some fame because people can't seem to realise that it's crap. They definitely passed the Turing test, while still being pretty much useless.
There are many other useless ML algorithms. Just because you don't like something doesn't mean it doesn't belong. ML has some good stuff and some bad stuff. The statement "ML works" doesn't mean anything. It's like saying "math works".
There have been many AI bubbles in the past as well, as well as slumps. Look up the term AI winter. Most AI algorithms turn out not really working except for a few niche applications. You are probably referring to these few as "ML works". Most AI projects fail, but some prevail. This goes for all tech though. So... tech works.
What Microsoft is doing is they are trying to cast a wide net to see if they hit one of the few actual good applications for LLMs. Most of them will fail but there might be one or two really successful products. Good for them to have that kind of capital to just haphazardly try new features everywhere.
No, they're not "all ML". ML is the whole package, not one part of the algorithm.
Obviously if you apply any tech badly it isn't magic. ML does what it's intended to, which is find the best model to approximate a specific phenomena. But when it's applied correctly to an appropriately scoped problem, it does a good job.
LLMs do not do a good job at anything but telling you what language looks like, and all the investment is people trying to apply them to things they fundamentally cannot do. They are not capable of anything that resembles reasoning in any way, and that's how the scam companies are pretending to use them.
They are all ML. I don't know how to convince you of this so I give up. Bye. I have a Master's degree in Machine Learning, btw.
No, they absolutely are not. You should go get your money back, because you very clearly don't know what you're talking about.
Machine learning is, by definition, targeting a single problem space. Using similar techniques to just shove any and all data at an algorithm and taking whatever dogshit gets spit out is categorically not the same thing.
We can do all those things without AI. Why do you care how fast it happens? If we could cure cancer twice as fast by grinding up baby animals would you do it?
Probably not the best to imply you want cancer treatment research to slow down simply because you don't like the tool used to do it. There's a lot of shit wrong with our current implementations of AI, but let's not completely throw the baby out with the bath, eh?
I dont want to slow down cancer research. I want to slow down AI research.
I care how fast it happens because I don't want it to slow down.
We can't just use the fear of death to justify any means to prevent it. If we found out we could live eternally but had to destroy other creatures or humans to do so, we would consider that to be too high a cost.
The cost for AI at the moment is just immoral. Even those who have found methods to deal with the costs, are still benefiting from calling it AI in the form of investments and marketing. Calling their work AI is worth money because of all of this fraudulent behavior.
If I started growing and producing my own organic abuse free heroin and selling it, it would still be immoral because I'm benefiting from the economy created by the illegal market. I'm participating in that market despite my efforts.
Ive said before that if these companies doing the ethical AI stuff want to stop being criticized for being part of this AI nonsense, feel free to call it something else. AI is overly broad and applied incorrectly all the time as it is anyways, and is mainly applied to things to draw money and interest that otherwise wouldnt exist.
Its a way to signal to investors that there is a profit incentive to be focused on here.
Sure, there are costs that are too high for anything.
This is the part where it breaks down though. There's nothing inherently immoral about AI. It's not the concept of AI you have problems with. It's the implementation. I hate a lot of the implementation, too. Shoehorning an AI into everything, using AI to justify a reduction in labor, that all sucks. The tool itself, though? Pretty fuckin awesome.
Are we comparing this to cancer research still? If so that's a bit of a WILD statement. It's pretty close to the COVID vaccine denial mentality - because it was made using something I don't like/fully understand, it must be bad.
Ok let's go back to drugs, then. If we were making your organic, free trade heroin, but called it beroin so that we're not piggybacking off the heroin market, we're good? No, that doesn't make sense. Heroin will fuck up someone's life regardless of what you call it, how it was produced, eetc.There's (virtually) no legitimate, useful application of heroin. Probably not one we'd ever see the production of broadly okayed.
Conversely, you've already agreed that there are ethical uses and applications of AI. It doesn't matter what the name is, it's the same technology. AI has become the term for this technology, just like heroin has become the term for that drug, and it doesn't matter what else you want to call it, everyone already knows what you mean. It doesn't matter what you call it, its uses are still the same. It's impact is still the same.
So yeah, if you just have a problem with, say, cancer researchers using AI, and would rather them use, idk, AGI or any of the other alternative names, I think you're missing the point.
I'm not saying they should do research at all, just take steps to separate yourselves from the awful practices practiced by the bug players right now.
We should be able to talk about advances in cancer research without having to have a discussion about how AI is going overall, including the shitty actors.
And, to be fair, most of the good projects you are defending do differentiate themselves very publicly about the differences and how they are more responsible. All I'm saying is that is a good thing. Companies should be scrambling to distance themselves from openAI, copilot, and whatever else the big tech companies have created.
yes. i love animals but if my kid has a cold and i knew a puppy's breathing caused it, I would drown that puppy myself. let alone finding a cure for fucking cancer.
that being said AI isn't doing that and even if it is I wouldn't trust the results.
Hey at least you owned the logical conclusion of your argument. I can respect that.
I do disagree, but I'm also vegan so thats probably why.
fair enough. i respect vegans.