I saw this counterpoint on Mastodon and I think it sums up things rather well: https://hachyderm.io/@chris_e_simpson/116925250364867462
Child labour is such an excellent comparison to use against the “if it is effective why not use it” argument.
comparing LLMs to child labor is summing it up well?
jfc
This has got to be the worst argument I’ve ever seen.
And watch the billionaires nod in agreement to more child labor lol
I really want to link this article here as a counterpoint - why “AI is just a tool” as a stance doesn’t quite cut it as far as well-founded opinions go: https://www.frank.computer/blog/2025/05/just-a-tool.html
My concern with arguments like this are what policies is this guy advocating for and who gets to define what a just system is. AI is clearly political, but it’s not at all clear who it’s gonna end up benefiting or harming in the long run. I strongly suspect it won’t be the billionaires because they have consistently failed to keep control of technology, and it’s clear this one is already getting beyond them – We have local LLMs.
Furthermore, large parts of what AI doing to society are how people are using it. The AI psychosis, lazy slop, alienation, or cheating in education – all of these are people using AI wrong. It’s not an inherent property of AI. Crucially, both people and institutions are already beginning to adapt, and I think we’re gonna be better off for it.
Like for AI psychosis, people should be developing the habit of not trusting things until they’ve thought it through for themselves – that’s huge for getting people to think critically. Similar for cheating in education, we’ve been complaining about education systems that teach the test for decades. Now, our education institutions are being forced to give kids actual projects and track their progress with those projects regardless of what tools they use. This is related to the AI slop – I’ve set up a series of prompts to have the AI critique my important writing. Rather than having the AI write for me, I’ve created an additional check for quality. I’d be shocked if people weren’t doing similar things for code and images. You can create more resistance for yourself to produce better works.
I’m not going to pretend to have a response to alienation, but I went months between talking to people well before AI, so how bad can it be. Plus, something about “emotional labor” and women not wanting you to vent/whine to them like they’re your mother. Idk.
There are environmental issues, but if you can get the LLM running on your own computer you’ll realize it’d not be very much of an impact if it weren’t for weird data centers. We should ban all data centers that aren’t powered with 100% renewables. I’m all for a policy like that, but I’d be very opposed to a blanket ban on all AI.
The development of AI also doesn’t bother me. Part of being a member of society is contributing to collective works. In my field (structural biology) we spent the last 55 years meticulously organizing data as an open reference for anyone and everyone to do whatever they could to make the world a better place, and when AI came along we were happy. The problem here is that it hasn’t yet been made to go both ways in LLMs, but weight distilling is quickly resolving that issue. I’d be fine with stripping AI of its IP, and forcing them to charge profit via ads or tokens if we really must run them remote. I don’t support the insane claims that they owe everyone and their brother for IP violations and the complete destruction of fair use. Honestly, I don’t want to strengthen intellectual property laws in general monopolies are stupid – set up a donate to get past the release threshold and then just release your work. Charge of consent update services or bug priority then release rarer free updates. It’s not that hard to not be a be a prick about it.
The economic issues are just same conflict between the rich and the poor we’ve always had. You should have unions. You should be retrained instead of fired. There should be quality standards. Rich people should stop trying to be exploitative enshitifiers.
Anyway, my point here is a lot of the people who are blanket Anti-AI are kinda missing the interesting politics that are actively developing in favor of an “change is bad” stance that’s crippling the public’s ability to take advantage of this transition period and advocate for things that’d actually help us.
i think most people here conflate ai with data centers. datat centers exist with out ai and you can use ai without data centers. local ai fills a lot of needs and in the next two years there will be more home hardware to run it. nvidia just announced a new consumer chip , coming out later this year, to run models at home.
in my eyes, the bubble has always been paying a company to use their ai. sure it has places for science and medicine but most of us dont use it for that.
so AI is a tool, how its used or implemented is the problem.
That was unnecessarily wordy to express only two complaints:
AI causes environmental damage and AI was created by scraping content that the AI companies didn’t own.
Did you read the whole thing? It’s about a lot more, such as corporate control, economic hardship, and alienation. It’s not a long read and you don’t have to care or agree with the author, but this is a disservice to their post.
This blog post is a wandering train of thought on the topic of what tools are and why it matters to be even slightly more mature in how we think about them.
That’s the second sentence, and it’s fairly clear that the author means to start the discussion of the titular topic, not to conclusively explore every ail of AI. Of which there are many, yet enumerated.
Some people call this “food for thought”; I would agree.
You can have a wandering train of thought without writing 2500 words repeating the same two ideas over and over and over.
He’s not wrong.
The solution is to make sure those LLM tools help maintainers instead of just causing them pain. There’s no question on that side.
We’re not forcing anybody to use it, but I will very loudly ignore people who try to argue against other people from using it.
And no, AI isn’t perfect. But Christ, anybody who points to the problems at AI had better be looking in the mirror and pointing at themselves at the same time.
Linus is, as usual, a pragmatist. And he’s right. All the bitching and gnashing of teeth over AI is ridiculous. If it produces good code then you should accept that code.
People will bitch and complain about ‘slop’ but the kernel team has processes in place to manage and review code from thousands of sources. They know what they’re doing.
Nobody serious is going to say “just accept AI created code without review!” That’s tech-bro BS. Quality software engineers know to adopt tools cautiously and deliberately.
If you’re going to argue people should accept then code (what people ought to do) then you’ve entered a moral arguement. People are free to argue about when the code is good as in useful but naturally that’s not as important as when the code is good as in morally good.
I value software freedom and copyleft licenses like Linux’s GPL for protecting it. While AI generated code cannot be copyrighted then it cannot be licensed under a free software license. You ought not accept “good” AI code.
AI generated code cannot be copyrighted
It can legally become proprietary
pick one
It’s funny how you pretend this is a ‘moral’ issue but then just end up with a copyright law conclusion.
While AI generated code cannot be copyrighted then it cannot be licensed under a free software license.
This is a concern - though the legal question is still a bit open on this. The question has been around how much human involvement there has been. A purely “create me a photo of a cat” prompt generating a picture of a cat has very little human involvement (fully AI generated).
A “generate some code to do this, no make it do it this way, rename these variables, etc.” prompt “conversation” may be treated differently by the law. We don’t know yet.
It makes no sense to say to someone what they “should do” without the part where it’s helpful to them flourishing or avoiding missery (i.e. morality). This can be expanded to include others in purely selfish terms.
There are no moral issues to AI if we ignore all negative ways it’s creation (and use) affects people. Typical AI-creator take advantage of others’ works on-mass while we’ve punished normal people harshly for far less infringement/social violations.
If the purpose of copyright is about encouraging human creativity then the AI generated elements which are clearly seperatable from human input should not be copyrightable? There can be creativity in writing promps but the resulting code output is not itself creative?
