curbstickle @ curbstickle @lemmy.dbzer0.com

Posts

10

Comments

1,537

Joined

1 yr. ago

Moderating

---

Great job!

For me (and not dinner tonight), two pumpkin squash, 1 regular sized cucumber and two gigantic ones, and some onion sitting on the table pulled this morning.

The garden is a jungle right now and I love it.

but if everyone used their own LLM, surely that would be even less efficient and more harmful than centralized data centres handling billions of queries (unless of course everyone is using sustainable energy sources).

Not really.

Those corporate LLMs are "everything and the kitchen sink", tuned LLMs are much more efficient.

Now as I've said I'm not doing image generation, but I do run quite a few models at home for various purposes. I also have an excessive amount of hardware for my home lab, work, and for general stuff at home.

And yet when I get my power bill with the comparison to my neighbors, I'm lower than average. And I dont have large PVs on the roof or anything, I just use efficient hardware and the rest of my home is specific energy efficient choices.

I do have some pv supplementing, but that's for some led lights I run for various purposes, connected to solar battery banks. We are going to move, so its not worth it to start a full PV setup here.

Now even without an efficient model, let's take a few more factors. Running a model at home its just responding to your query, no additional data or other processing required.

The corpo stuff though, they are looking at your identity and data, evaluating the question related to advertising and what can be sold to you, or who your information can best be sold to. This sort of stuff is adding on a ton of unnecessary garbage thats burning up cycles for.... No reason related to the LLM or other tooling, purely as a profit center at the expense of your privacy.

I pretty staunchly recommend against using corporate owned and run AI, it is detrimental to the environment. As are cars. As are large industrial efforts that sacrifice the environment (often ignoring regulations since the fines are lower than the profit they will make) for the sake of "line go up!"

Capitalism is really bad for the environment.

Don't know what to tell you. If you don't see the issue with taking another person's labor and exploiting it for your own capital gain, then I don't know why you're even leftist. You sound more like an opportunist at that point.

So capitalism, what I've been calling the problem this whole time?

Cool.

Corporations are the problem because they are exploiting and stealing other people's generated labor for capital gain.

There it is!

If someone else is doing that, it's still bad, because that's how corporations are started.

Boy oh boy, I had no idea people profited from their posts on Lemmy. Clearly I'm missing out!

You're right, capitalism is a problem. Abolishing capitalism is more important than fighting progress in AI development.

Ding ding ding!

But we aren't doing that. We aren't there.

Some are. Why aren't you focused on helping those folks rather than complain about a symptom of the problem?

So AI's current existence and commercial implementation is a net negative to society at large.

Just the commercial use, and how its been done.

How does that, in any way, relate to people making images and posting them to Lemmy? That some of them used a model that contained other peoples works? In what possible way does that relate to people posting their generated images to Lemmy?

example thats just capitalism as the problem

And yeah, I'm not generally that worried about people with their own LLMs or whatever. But they're not exactly free tools. Not everyone has equal access to them.

You can download them and use them, there are a ton of resources out there, both with and without the materials you're concerned a out, freely available and shared.

There is even a completely free AI horde available right here, with peoples donated resources. Thats about as accessible as it can be made.

More stuff where capitalism is the problem

So... what's your point? Fuck trying to fix things, everyone just do whatever you want, it's the end of days, hopefully revolution comes on its own?

My point is to focus on the actual problem, not be distracted by the latest method of exploitation. There is always another method, and focusing on just what's in front let's more line up behind it.

Original point: some people (didn't say me) think it's immoral.

And it always comes to this immorality being based in this point:

Secondary point: IP laws aren't inherently immoral, they conceptually exist to protect laborers from having their labor exploited by people with more existing capital.

So.... Capitalism

They aren't even good in their current implementation, but believing in some form of IP law under capitalism is essentially a socialist policy.

..... Wut?

Third point: When most people think of AI, they think of corporate use.

Yup.

Literally no one gives a flying fuck about your personal models trained on your personal data

Comments made across this post, as well as the mass down votes of AI communities, would absolutely disagree with you.

Since thats the entire subject here, seems like misplaced anger dont you think?

Fourth point: It's more important to worry about where we are at than where we want to be. And look around. AI is a problem, and it needs regulation, and regulation of it includes protecting ACTUAL artists.

I agree that you can't ignore things happening right now, but AI has not decimated the workforce. I even gave examples of people who do contracted creative work who explicitly said AI got them work because the work needed to be redone.

Ask any programmer how much they would trust an ai generated application. Let me know when you find one who says anything other than a wild laugh (that doesnt call themselves a vibe coder, pro tip, that is not a programmer).

And then let me know when you can tell me how any of this relates to people posting images to Lemmy.

No, I'm pretty comfortable where I'm at.

I think intellectual property is a functional element of capitalism, and capitalism is a problem. I support people directly, and could not care less about anyone downloading content. I also support the local art scene, through events and direct purchase of works.

That doesn't mean that I think intellectual property (as a product) is sensible in any way.

That said, all my models are personally trained on my own data, and maybe a public domain based model for basic language to put context to a detected issue. That doesnt mean I'm going to put down folks who use a local model trained on some degree of copyright works, because no matter what, those users aren't the problem. Corporations are.

Because capitalism is a problem.

I think it's a more fair statement to say almost anyone hating on AI is usually referring to corporate AI in the first place.

I'd say its more accurate that the people who constantly hate on AI are ignorant of the tools and their use, and most often their position is nothing more than "capitalism is the problem" (see above) whether they realize it or not.

And some will even go on to defend capitalism, which makes it even weirder to me.

And considering some of the comments across this post, I'd say my perception is pretty accurate.

Personally I'm for it, I think it has some approaches that are more interesting (topics for example).

So as long as its not a pain to run, I'm into it and would try it out for sure.

I'm shocked. Shocked, I tell you.

"Fixing capitalism" is just ending capitalism, the problems are inherent in its design.

In any case, you seem to be under the impression that all AI is just chatgpt, copilot, Gemini, grok, meta, etc. This is not the case. There are models trained entirely on peoples own work/material, models trained only on public domain materials, etc. Its just software and data, and the issue you are taking is - from your example - a capitalism problem, and has nothing to do with the software.

So what you're saying now reads no differently than "We have to fix capitalism before we can advocate for Linux!"

Would that make sense as a statement to you?

At work, a CTO for a firm we were designing a solution for got upset because hardware was in customs, and tariffs needed to be paid.

He didn't understand why the manufacturer wasn't paying, and it had to be pointed out (by someone from the other company bringing this hardware) that tariffs are a tax on importing, not exporting, and for every single contract they have ever had, tariffs are paid by the client.

This was not a small company, either. We're talking about a fairly large firm in finance. And the CTO didn't understand how tariffs work.

We're fucked.

I dont advocate for corporate AI.

I'd also mention most users on db0 are far more likely to be using fully open models and models they've trained themselves (which is what I do, mostly log/error eval stuff).

That said, "fixing" AI will solve nothing, because capitalism is going to find yet another way to screw them over. Banning AI tomorrow isn't going to provide job security or a stable income, it wasn't before and it isn't now.

I work with a lot of creatives, and so many are contract based and struggle between them. Several of them have been finding work cleaning up (as in, creating new) materials that were AI generated and look terrible.

So again, the issue is capitalism, and AI is just the most recent conversation piece. AI isn't the root of the problem, nor is the problem "solved" without it.

Yup, was getting an uber (I can't drive for very long or the pain starts), he was chatty and excited about Trump.

Asked what he thought about the anti-immigrant talk (oh obviously thats the bad ones!), and the fact that there was a pandemic and prices skyrocketed (Biden's fault even though that all started before he was in office), etc. Paraphrased obviously.

I don't know what can possibly be said to these folks to have them understand, but as a result, we have... Well all this current nonsense.

Thats a capitalism problem, not an AI problem. AI is just the most recent example.

They shouldn't have to rely on maybe getting paid so they can have food, shelter, and other basic things for living.

This is one of those cases where an algorithm carefully trained on only relevant data can have value.

Hopefully more people learn that this is the important part.

It becomes nonsense when you just feed it everything and the kitchen sink. A well trained model works.

8 months ago:

Literally in a car with an immigrant right now who voted Trump because things were cheaper before covid.
FFS.

I'm curious what that guy thinks today. Or if he's still in the US.

So brigading small communities can be easier?

The 1yr old and 2 yr old were left alone in a car, for 90 150 minutes.

This was negligience long before a child died.

Edit - misread the times, its so much worse.

Thats what I thought.

Ive seen the "We have salt beads!" one recently, but none with the plastic microbeads.

By the opinion of a rando on the internet?

No, of course not.

Nah, just have no interest in people who behave like babies.

Your respect means nothing to me in any way, so... enjoy not having any for me I guess?

My mistake then, it read (to me) like you were saying you down voted once, in each of the communities you were banned from.

Then all I could guess was timing. Multiple communities might depend on what client they are using, but in any case a single downvote to a ban rings as mod silliness to me, I agree.