If LLMs used very little RAM and processing power, was light weight even for Mythos or Sol, would people be as opposed? Like no data centers, no memory shortages, like it was able to run on a calculator with the same performance.
The resources used are but one of the problems with the modern uses of AI. Others include:
- Copyright and privacy. AI companies use copyrighted content and private data to train their models. Many companies have terms or user agreements that give them permission to use or sell any data on their systems. They collect data where they can, without regard for ownership, copyright or privacy.
- Fakes. We cannot trust what we see or read. Of course fake news and misinformation has been a problem for years, but genAI has turbocharged it, and now even video is becoming unreliable. We cannot trust anything anymore. Even appreciation of real art suffers because artists get accused of using AI even when it was hand made. Although the impact on politics is the biggest threat.
- Companies force employees to use it, even when employees don’t want it and it hurts the quality of their work. AI is often used as an excuse to fire people.
- The AI bubble dominates the economy. This cannot end well.
There are also legitimate and useful ways to use AI, and many of them could be very valuable, but that’s not what the industry or society is focusing on right now.
AI would probably be more acceptable if all of these issues were addressed.
Absolutely not, I hope.
The impact on the minds and the constant push for the thing for all the wrong reasons are enough for me to keep hating it.
And they always add it to everything, even where is not needed/wanted. It is a cool tech marketed and implemented in a bad way.
Where I think the bulk of the hate comes from is less the tech, but more the political and business elite pushing this tech.
These people are going full mask off trying to delete those they deem the lower class, so much so they are blinded in how much they require people outside the LLMs to maintain society.
LLMs as they are now, are basically the Golden Calf scenario all over again; it is ungrateful people build a statue to claim superiority and mock that which put them in good places.
Yes. Cause the resource intensity isn’t the main problem people have with it.
It’s the theft.
It will always be resource intensive since companies are in a dick measuring contest of who can sink the most resources into it. If it was lighter they word simply do more of it
Slop is still slop, and if it was cheaper and less intensive in those ways it would simply be producing way more slop.
The main problem for me is that it’s so confident in it’s tone, like that idiot friend who just makes stuff up, but totally believes it. So there’s no clue (like you would get with a normal person) that the thing has no clue if what it just wrote is true or not. I could bear it more if it said “I’m not sure, but…” But then if did that then none of the idiot CEOs would buy it, so here we are…
It is functioning in similar vain to idol statues placating and validating the idolmaker. In fact, that’s my go to term for it because it’s ridiculous for they practically deify this tech.
These ‘AIs’ are still dependant on human input and human information (not knowledge) and the fact they follow a business model and not academical, means they are always seeking to earn a customer’s approval as opposed to teaching a student what they need to know.
Would be even more. The main harm of AI comes from malicious actors being able to ruin the internet forever with practically undetectable bot spam costing cents to run. Also taking jobs in art, where automation was never possible as efficiency is not what art is valued for.
it’s one segment of the problem but not all of it by far
Well if that was the case tech bros would just strap 1000000 calculators together, call it mega AI and proceed to claim it is the next best thing that ever happened to humanity and billionaires would fund them. It is these opportunistic vultures that enshittify everything and make them anti-human.
If the AI today always proceeded with the mind set of openness and locality it could still produce very useful tools for a fraction of the environmental impact it causes now but they instead, as always, have chosen thr most aggressive possible method because it can be hyped more.
LLM the technology is interesting and impressive. the way companies are shoving it into everything and trying to sell it as a replacement for employees is awful. The way the AI companies are behaving is awful.
AI is great! it just needs a little work on the IP theft, the environmental impact, the dopamine addiction, the cornering of the component market, the reliability, the hidden system prompts, the non-determinism, the push to have everything in the cloud, the enshittification, the empowerment of the billionaires, the reduction in user’s cognitive abilities, the lack of real-world use cases, the reliability, the spying on humans intimate thoughts, the overblown marketing, the bubble impact on the world economy, and the reliability. Other than that, it’s just fine! Trust me, bro!
The same can be said about LITERALLY every technology ever made. Everything you listed is real, problems that need to be addressed, and the companies responsible for exacerbating them held accountable.
A technologies worst implementations are not representative of its usefulness.
Unless you want to ban the entire field of biology as well for the atrocities committed associated with it.
Yeah transformer models are cool.
But you cannot separate the industry from the tech in this case. The reason it’s so bad is because it’s being misused and it’s everywhere, even where it shouldn’t be.
No, that’s not true. Universities have been making their own models. Private individuals have been making their own models. The most famous ones are, of course, purely corporate. But claiming that the tech is inseparable from the corporation is demonstrably incorrect.
Evo 2 was recently used in a revolutionary antibiotic research paper, which is entirely open source and trained and public data. Is that also corporate?
It’s great! Except for the several teen suicides
The problem is it’s designed to bullshit you. Why would I trust a foundational problem?
It’s also designed to eliminate your job and leave you for dead.
Yeah except the parasitic sociopathic class is already trying to find ways of getting rid of most of us from the for profit margins. Every employee takes money from the profit.
Any good thing is ruined once it’s co-opted by capitalism.
AI and LLMs (In my very very personal and completely subjective opinion) is the coolest shit humanity has ever invented and under the right system would’ve been unanimously celebrated.
But noooooooo, capitalism had to go and fucking turn a miraculous tool for human creativity and knowledge into a relentless driver of layoffs, corporate slop and copyright theft because some greedy fucks want more money for themselves than they know what to do with.
But, well, that’s just, like, my opinion, man.Yes, because of the way its used, the way its presented, and how bad it is.
Different people would hate it for different reasons tho
The most common use for AI seems to be Nazi/Fascist/Republican propaganda. Basically anywhere you go is an endless number of bots pushing to defend neo-nazis, tell people that Trump totally isnt a pedophile, make posts about “If he had just complied he would have lived” etc etc etc.
The subset of people who recognize this, would hate AI because its really only good as a propaganda mill.
The next set would be because of the way its being pushed to replace workers, AI doesnt actually make the remaining workers jobs easier, and it doesnt do work more efficiently, its just another excuse to cut headcounts, make a shittier service and charge more for it.
Finally is because its A) Very often wrong and B) Being used to replace actual critical thinking.
Its a growing problem, especially in academia. Students are coming into college or high school with absolutely no idea how to do actual research, how to write a paper in a logical way, how to follow basic logic chains etc. And there is so much bad information being produced by AI now that its becoming harder and harder to find actual good information. For every doctor telling you to vaccinate kids, there are a dozen AI driven videos that have some person being confidently wrong about everything to do with vaccines.
Are there still students who didnt cheat their way through high school with AI? yes.
But they are becoming more and more rare.
And given how wrong AI usually is (AI is very good at being confidently wrong), this means that we have a whole generation growing up who cant tell the difference between “Measles can kill you if you dont get vaccinated” and “Shooting gay people is a fun weekend activity”
It would probably be hated even more as all of those points would be magnitudes higher in volume than they are today. It’s what is wanted by corporations, a cheap surveillance tool that closely monitors everyone to keep everyone in line. There are multiple dimensions to this issue. I think AI should have stayed in R&D for a while longer. It is so half baked, but unfortunately trusted as an “expert” by people who may not be familiar with hunting down primary trustworthy sources. Though that last point is nothing new. I think it should have never been used for things like scanning private/personal files or to accuse people of crimes they didn’t commit. These companies are buying their own bullshit and are trusting/betting on AI to reduce the most costly component of their company, the labor force. It’s all just smoke and mirrors though, always has been. It’s not like these things are gods among us, but they are sold that way. In the end, not matter what, with the current state of AI or if they were cheaper to use computationally, the bubble will burst. At least when that happens, maybe some of the left overs can be repurposed for actual useful things for society, but I don’t have my hopes up at the moment.
There are 3 reasons people hate AI right now:
- Being used poorly and mostly resulting in slop, or malicious/unethical/deadly uses, and psychosis
- Being forced down the public’s throat and put into places where it adds no value, job losses, quality drops, plagiarism
- Datacenter/environmental concerns, part prices
So switching to local AI fixes problem 3 (mostly). But we also need the hype bubble to pop so number 2 goes away. Unfortunately the first problem will be hardest to fix and right now requires the public backlash as new norms are being formed and bad actors need to be called out.
That said, I think we’re about 1 to 3 years away from the bubble popping and everything starting to balance out to where AI is treated more like just a tool rather than a genie in a box. The hype will fade out and one or two of the big players will die, but at this point I don’t think generative AI use is going away
