Yes, but it is very important to distinguish between LLMs, which are amoral math, and OpenAI et al, which are corporations run by fascists that offer LLM-based public surveillance / disinformation platforms. There is no technology on earth that cannot be bent to malign purpose if malicious people are inclined to do so. LLMs are just the popular useful technology du jour.

Even most of of the local and "open" models are exploiting the work of others by training off unethically sourced data without even so much as consent.

There is such a thing as ethical AI, but damn near zero people are doing it that way.

I got called unethical today for using a small-scale locally hosted open model trained on Wikipedia and Stack Exchange. A LOT of the anti-AI people are universalists that have decided carte blanche that it is an evil technology and everyone who uses it is bad.

The torment nexus isn't a discrete thing. You can't make thumbscrews without screws. Are the screws evil? If not, when does the composition of components become evil? I don't think LLMs are evil for the same reason I don't think thumbscrews are evil in and of themselves. Thumbscrews, like LLMs, lack the capacity for evil. It's the people who turn technology against their fellow humans that I hold as my enemy, not the trappings of their mischief. Remove them from the equation, and the torment nexus becomes so much harmless scrap, ready to be beaten into plowshares.

surveillance networks are made of computers and sensors, which are made of chips and wires. I don't think chips and wires are fascist, I don't think sensors and computers are fascist. I think the purpose to which they are put is the fascist part, and the purpose is pure human.

I wonder if it isn't at least partially a psy op. This whole business of giving anyone who even looks at an LLM a scarlet letter, and refusing to participate in any critical evaluation of LLMs or other AI tools, plays right into the hands of those who would use LLMs to pollute, isolate, and destroy the free internet. The willfully ignorant will be the easiest to trap.

If you can read this: beware the day that internet discussions of mass manipulation go quiet. That silence is the death knell of freedom.

No, but I believe machine translation is now powered mainly by LLMs, which is a damn useful application of the technology.

I was surprised to learn recently that a few of my friends pay for Claude to help with personal projects and things like that. This particular AI tech is quite new so it's surely too early to declare axiomatically that it has no uses. That sounds like an emotional reaction to me.

No it isn't. That's the distinction that is made when arguments arise, but the claims really are spawned indiscriminately.

You need to understand what you're complaining about before you can name it, and most people don't understand much of anything about modern AI and LLMs.

I've been an AI researcher for 12 years. Generative AI also has many applications that tangibly improve the world. For example, I'm using diffusion models for super resolution of surface wind fields to forecast wildfire propagation. It's not "just" generative AI that's the problem. The exact same transformers or CNNs used for Capital to try to finally triumph over Labor can also be used to do all kinds of amazing things.

My point is that the incentive structure is fucked. It's just more apparent with a powerful and widely known technology like "AI" these days.

And yet it’s LLMs doing these things, not only ML. People don’t cure cancer either, but the things that get close (eg. AAV viruses) are currently being done with LLMs.

AI means, and has always meant, artificial means of doing what before computers seemed only possible with real, human intelligence. Another good way of looking at it is algorithms for problems for which it's not practical to explicitly write an algorithm to solve.

It's meant this since Turing. That's a remarkable foresight if it's "purposefully poisoning the well."

Nah. The math is all the same, the models can be the same, it's all just the application. We don't like AI applied to bullshit capitalist profit driven goals, but those exact same models are being used to tangibly improve the world in other applications.

It's like saying hammers are evil because people can murder people with hammers, which is ignoring the entire industry of construction workers. It's just in the case of AI, it's way more powerful than a hammer, so the implications are much larger depending on the application.

Respectfully, you have no idea what you're talking about. I've been an AI researcher for 12 years now. I work primarily in climate and environment, and right now I'm using AI for super resolution of wind field predictions, to better forecast wildfire spread. Prior to this project I was using AI and remote sensing to monitor and predict wheat crop health in northern India to feed into various food security monitoring systems (e.g. FEWSNET). There are millions of AI use cases that tangibly improve the world.

At the same time, those same underlying principles of AI are being used to develop tools to better surveil people, control them, and profit off of them so the tech fascists can gain even more control. Both things can be true. It's the billionaires who are funding these massive investments that are evil. It's always the billionaires.

Steel can be used to construct buildings and can also be used to construct weapons. You wouldn't say steel is evil right?

Brother you can run that shit local and all of your claims have so many asterisks attached that they're essentially pointless. Everything you claimed is entirely true of base computers no ai. Hell even just internet access has been shown to have all the same negative effects. And I'm defending it because I'm not stupid and it's a useful technology. Run your shit local. The tech bro goons don't own the technology.

That's a gross overreach and obviously wrong. Selection bias at its finest. The people who act as you have described are loud and easy to spot. The people that treat it like a tool don't go around telling everyone about it. You're conflating improper use and corporate abuse with ANY use. By that logic we should outlaw X-rays entirely, since when they were first discovered they were used fuckin' EVERYWHERE, and quite a lot of people got sick because of it.

Humans are bad at moderation. When a shiny new toy is presented, we obsess to our own detriment. Over time, a balance is found. This has happened over, and over, and over, and over.

And the person using it.

Please replace retarded with another word so that I don't have to remove your comment for ableism (it has been reported.) Thanks!

The meme presupposes the usefulness of AI, which is why everyone's taking it as a given. If it isn't useful, you don't need to use normative arguments to stop people from using it, people will just naturally never find an opportunity to do so.

Also if you really can't think of any ways to use a chatbot to advance your political goals you need to think harder. It just shows a lack of adaptability.

Always glad to help. And yes, the cringe was palpable and something needed to be said.

That is a very good point, Master Wayne. Thank you for calling me out. Going forward, I will try to avoid such mistakes.

Fighting local models is playing right into enshittification, though.

They WANT people to be scared of "AI." They want people to keep talking about it as this existential anthropomorphized danger, not a tool theyve enshittified.

Because the thing that keeps Altman/Modi up at night is the proliferation of trifty, dirt cheap open weights models. Ones presented as tools, not intelligence. And every bit people rail against them is more ammunition to regulate those open models away, so they have a monopoly.

True, it's just a tool and should be used where it best provides a real benefit. That's essentially the core of my argument.

That's one use of neural networks we do need to react to accordingly as a populace but it's not the only use of those algorithms. I gave the example of alpha fold which has essentially solved the very hard problem of protein folding unlocking the door to developing cures for many diseases. The issue isn't the tech, the issue is in some cases how it is being used. Think of nuclear power. We could have functionally limitless and clean power or we can end life on earth. The problem isn't nuclear power, the problem is the actions of people.

Yeah, could very well be. The "do absolutely everything with AI no matter what" idea is a circlejerk, too. (Just in very different spaces from this one)

On my project right now, I'm doing everything by hand. It just seems unlikely hallucinate-and-cull would be any faster at producing useful and well-architected networking logic. On the other hand, when it comes time to write a UI, spinning something up is probably worth it.