Your strawman arguments bore me. No you muppet, I KNOW humans can spread misinformation, and LLMs make it so that it’s 1000x easier, and allow corporations to do so through their product without any accountability.

What strawmen? I’m just pointing out how your arguments against LLMs aren’t unique to them, and can be used by humans, and other tools, as well. Do you think Zines, Newspapers, Podcasts are bad? because they make it easier for corporations to spread misinformattions 1000x easier without any accountability.

Flat out no they dont. They use electricity to make pictures in the style of human art. Because all they can do is copy the effort of humans that have unwillingly been used as training data.

We have a word for “pictures in the style of human art.” We call it: Art.

I don’t see how machines making art is different from humans making art. If you can’t tell which one is done by a machine, and which one is done by a human, is it any less “art” ?

Because all they can do is copy the effort of humans that have unwillingly been used as training data.

Once again, that’s not inherent to LLMs. That’s how some of them were chosen to be built, but many of them are built with data that was willingly given to be trained on. If a human learned to create art by copying stuff that other humans didn’t permit them to copy, do you think they can’t create art?

Yes, your smartest friends aren’t throwing shit at the wall to see what sticks, theyre using their combined knowledge and experience to discern an accurate answer to a problem and if they don’t have the answers they can use that knowledge to find the answer efficiently. LLMs will guess the right words to say, then tell you they have the answer but immediately fall back the second its told that the answer was wrong.

…theyre using their combined knowledge and experience to discern an accurate answer to a problem and if they don’t have the answers they can use that knowledge to find the answer efficiently…

That sounds an awful lot like probabilistic modeling.

LLMs will guess the right words to say, then tell you they have the answer but immediately fall back the second its told that the answer was wrong.

Many humans do this too.

LLMs can’t do this. Yes, you are correct, and that is entirely my point. Why do you have one standard for humans, and another for LLMs? It seems to me that you should judge them by the same standard.

They are neither cheap nor accurate, they can merely use their training data to mimic human output.

They are cheap. I’ve used them plenty to do these things, and it cost me much less than a human. Humans are also not accurate.

These things all require humans that have trained to use these tools to perform their task accurately.

And LLMs require “humans that have trained to use these tools to perform their task accurately.” How are those tools any different?

I’m done with you, keep shilling for corporations to carry on ruining the planet for the sake of profits, politicians and the richest people alive all you like.

When did I ever do that? Can you point to a single instance? I never supported corporations, nor environmental destruction, nor politicians, nor the rich. I’m generally against all of those.

You seem angry about problems in our society, angers that I share: Wealth disparity, elite non-accountability, workers being mistreated, the rich and powerful leeching from society, etc. But I don’t see how LLMs are the cause of any of those. All of these things exist because of capitalism, imperialism, corruption, etc.

How about human thought. How about the mentap state of the people that use it. How about the general state of both online discourse and in person interaction. How about the level of misinformation being spread either intentionally or by the machine that tells people they’re right all the time providing confirmation bias to the people that are wrong.

So you think that humans can’t cause bad mentals states to other humans? You think animals can’t do that? You think humans can’t spread mis-information, or biases? Or do you think we should get rid of all humans too?

Tell me you dont understand how intelligence works without telling me you dont understand how intelligence works.

Do you have an actual point? or are you just trying to mimic things that you don’t understand, in a sad attempt to “own” me?

Alright, what I want you to do is have an LLM tell you how to dismantle diagnose and fix a fault in a car, without making mistakes and without breaking anything in the process.

The vast majority of humans can’t do that. I also don’t see how that has any relation to the statement you were responding to.

What part of having a tool and skills takes away from humans and puts the benefit in the hands of a corporation in the same way an LLM made by a corporation takes away from those skilled humans?

Computers. Printers. Tractors. Cars. Websites. Courts. Guns. Many things do that, yet I still see you only singling out LLMs.

Video games USE electricity to produce art and provide a source of entertainment. Cars USE epectricity to transport humans and cargo from one place to another. Spreadsheets USE electricity to store data from sources such as shops, businesses and people that want to organise information, grow lights USE electricity to provide light for plants to grow.

LLMs USE electricity to produce art and provide a source of entertainment. LLMs USE electricity to store data from sources such as shops, businesses and people that want to organise information. LLMs USE electricy to transport information from one place to another.

LLMs waste epectricity by spending time doing something that could easily be done by a human being at a fraction of the cost. WHILE DOING IT INACCURATELY, forcing a human to read through the outputs to identify those inaccuracies, wasting time and electricity by increasing the work required to get an output that a human could provide.

So humans can analyise large lines of code in a matter of seconds for cheap without doing it inacurrately? Humans can create dozens of pieces of art cheaply, in a matter of minutes without any inaccuracies? Humans can generate dozens of stories, cheaply, and in a matter of minutes with no mistakes?

Interesting, because I’ve never met humans who can do that.

All of these HELP people to do work better or provide avenues for skilled workers to do their job better.

But they are also “…things which could easily be done by a human, taking away from actual people and putting the benefit of that work into the hands of corporations”.

They fit your criteria perfectly.

Language models GUESS what word they should say next and fake being smart…

Tell me you don’t know how LLMs work, without telling me that you don’t know how LLMs work. Guessing is a very powerful tool, and the smartest people I know spend a lot of their time “guessing”.

…while wasting electricity and time by making a human into a glorified spellchecker, because any technical work they’re put to is delayed by adding an extra layer where a professional with the experience and expertise needs to check every output for accuracy.

Video games “waste” electricity. Cars “waste” electricity. Spreadsheets “waste” electricity. Grow-lights “waste” electricity. I don’t see how LLMs are all that different.

I have to check the output my co-workers do. Does that make me a “glorified spellchecker” ?

I get that you don’t like LLMs, but if you want to convince me of anything, you’ll either have to make your values consistent, or come up with something that LLMs damage that is unique to LLMs, and inherently part of their function.

I think people are upset by the legitimate problems that they see, but I think they typically mis-diagnose the cause of the problems.

People are scared to lose their jobs, and their livelihood, for good reason. Capitalism sucks, and it ties the two together.

I’m pro automation, because I want people to lose their jobs. What I don’t want is for them to lose their livelihood. I think people working on jobs that they don’t like just to pay the bills is dumb, but people working on jobs that they enjoy is awesome.

Who needs to process the camera data? Again, it is a very different technology from all of the comparisons because it is literally taking out the need for large bodies of human oversight. Data collection vs data processing.

I could say the same thing about spreadsheets. Is spreadsheet software, or databases bad?

A camera does not decide who lives and who dies. An LLM can (and has) decided who lives and who dies and can carry out the execution orders itself.

That’s a choice that is being made. That’s not the point of LLMs, it’s one of the many things that LLMs are being used for.

But I guess cameras aren’t starving people of water and power and a required prerequisite to their use at scale.

The same is also true of LLMs. Capitalism is doing that. We can make LLMs much more efficiently, but we choose not to, due to capitalism.

But you’re right. LLMs aren’t evil at all, they are just a tool. The next defining tool that fascists now need to subjugate everyone and eliminate resistance.

I’m against their use by facists, just like I’m against facists using cars, or bulldozers, or guns, or wheat. I just don’t see how LLMs are any different.

Letters can be delivered by a human. Do you think we should only send letters, intead of sending texts? What about cars, and trains? They take jobs away from people, should we abandon those? What about copiers, or printers? What about power tools? or water pumps?

Those things all help corporations, and non-corporations too. I don’t see how LLM’s are any different. Should we abandon all automation to keep people employed?

There really isn’t any point in wasting any more time arguing with them, they weren’t arguing in good faith and aren’t going to.

Yeah, I figured that out, and had already blocked them. But thx!

If LLM’s don’t have a well-defined problem to solve, how does one misuse it, if they are not for “direct calculations” you keep talking about, what are LLM’s for?

They are for statistical analysis and probabilistic modeling. They give general responses, based on statistical inference.

I’ve given you a few problems that aren’t capitalism specific, I’m not going to recount them, if i was more rude I’d suggest getting an LLM to fetch a few bullet points, but it’s more healthy to just take a few more minutes to re-read what i just took time out of my day for you all to read.

It’s always amazing to me when people accuse me of doing what they themselves are doing. I read what you wrote, but you aparently didn’t read what I wrote. I never claimed that you only gave capitalism specific problems. I’m not even sure where you got that from.

And the reason you don’t see me blaming other tools is just because this was your only interaction with me, I’m learning software engineering, because i like the subject, and in software almost all we do is complain about tools, the documentation for said tools, the setup required to use a tool and the like, critiquing work is part of what makes an engineer, that’s how we get better, usually though, i don’t need to argue that something is not even a tool.

I don’t understand what you mean by this. I explicitly pointed out tools: scrolls, waterwheels, hammers, etc as examples of tools that can be misused to be anti-worker, and hierarchical, but aren’t inherently so. Are you saying that you do think that they are “inherently anti-worker and hierarchical” ?

Please explain it to me, because I don’t see how that can be true.

This applies to cameras as well. Cameras help fascists, fascists help target victims, cameras help track people, cameras help report where people are and what they are doing. Cameras make us easier to control.

Do you think all cameras are bad? or are they just a tool that is being mis-used by the fascists?

Lol, the person who refuses to look at the data, accuses someone else to not look at the data despite evidently having no clue at all.

Yes, your self-awareness is impressive.

Point me to a single open-weight model on that list that isn’t a tiny toy model. Do it!

All of them.

LoL, the person who refuses to look at data, is accusing me of trolling them. Great success.

Yes.

They are open-weight, and I don’t see what being a toy model has to do with anything.

Many such cases. I see this all the time when it comes to genocide, or the DNC, or healthcare, or economics, etc.

People can be frustrating

…and are pretty aware of what is available and possible with them.

Yet, when presented with data that contradicts your point, you sure don’t seem to understand the basics of how they work.

But every time I point out the clear technical and moral issues with them…

Then why don’t you try doing that here? All you did was ignore data that was counter to your claim.

…that they clearly didn’t bother to even look into themselves and then ignores everything that contradicts their opinion and beliefs…

Oh, you mean like you just did? curiouser and couriouser

Did you? Because based on the post timestamps, you spent literally 5 minutes looking at that list, which doesn’t seem like enough time to look through or test any of those models.

All of your responses here are either wrong, or keep side-stepping my points: Yes, LLMs can give sources, and often do. Yes, people do shitty things with LLMs, but how is that inherent to the LLM? Yes, LLMs don’t have a well-defined problem, and can’t do direct calculations, that’s not what they are for.

I still fail to see how any of that makes LLM’s inherently bad. The things you blame LLM’s for are social problems, or capitalism problems, or something similar. I don’t see you blaming other tools for the problems they create when they are mis-used.

So you admittedly didn’t even look at the models, nor their uses, but have determined that they don’t match your criteria?

If you aren’t going to accept a response just because you don’t like that it contradicts what you want to believe, I’m not going to waste my time engaging with you.

If LLMs were only made by apartheid supporing corporations, you might have a point. But many different people and groups create LLM’s for all kinds of things: Statistical analysis, generating stories and characters, coding assistance, design assistance, process assistance, language translation, rubber ducking, debate assistance, running D&D games and many more.

You want me to do a basic internet search for you?

Fine: https://www.fairlytrained.org/certified-models

First, we have skilled labor for a reason. Some people dedicated years of their life to perfect a skill or a trade, and they are trustworthy experts with experience in the matter. It would be unfair for those people if they got replaced by a sloppier version of themselves, after spending so much time and effort into their skills.

Is it “unfair” to bakers that I make my own bread at home with a bread machine? What about using a calculator? Or a dishwashing machine?

Second, AI was built on stolen labor. Even if you use a self hosted model, running on old hardware that was thrown away, using solar and wind energy, its usage is still morally wrong.

Well, it’s not all built on stolen labor. A lot of it is built on data that was given away freely. You also didn’t really provide a reason why it would still be morally wrong.

On top of that, even if the models are used locally (and even if they are tuned locally), the initial training happens on massive datacenters that consume lots of resources. And what you get isn’t the first attempt at making a model. For every successful model out there, thousands of other models were trained and abandoned.

They don’t have to use massive resources, we just choose to do them that way because of capitalism. Cities use massive amounts of resources, and many of them aren’t successfull on the first try. Should we stop building cities?

The rest of your post is more of a complaint about capitalism, and misunderstandings of how we choose to make AI models, not how they inherently have to be made.