There is no ethical use of "AI".
@Gargron AI itself is unethical because it was developed using stolen data.
@Gargron I'm honestly digging the idea of local LLMs. I hate AI in the hands of techbros and those in charge of it. But if we can get a LLM that can work off a phone without killing it's resources it can be a great tool.
I agree with you. But it's ubiquitous. Can't do autocomplete without AI. Can't use translation tools. Can't escape. I've tried. I avoid it to the best of my ability.
Capitalism is also unethical by design. Can't escape that either.
@ryan You are on a domain called "artificial ryan" with an AI-generated slop profile picture and header, the audacity here 😂
@Gargron You are my favorite mirror-universe techbro.
@Gargron 'ethical' is less important nowadays 😔 And I'm not talking about the Tech Bros. Average people who are aware of the implications still use Gen AI tools.
@Gargron So is the “Translate” link that appears in many toots on the Mastodon web app not a use of an LLM?
@Gargron Please correct me if I’m wrong:
At the discretion of the instance admin, the translation is done either by DeepL or LibreTranaslate.
The former is LLM-based:
https://support.deepl.com/hc/en-us/articles/14241705319580-About-DeepL-s-language-model
The latter? What I’ve read says that they offer two options, one LLM-based and the other is transformer-based without being an LLM. That granted, I have a hard time seeing an obvious *ethical* difference between LibreTranslate and DeepL. Does LibreTranslate promise that all its training data is used with permission?
@Gargron Out of curiosity: what would it take for the use of machine learning to be ethical in even the slightest sense of the word? When it comes to accessibility, I see a lot of ethical uses for AI. Real-world tools and human help fall short for my wife, who is deafblind. AI provides literal 24/7 support for her. This heavily influences my stance on the topic, even though I don't fully disagree with you.
I copied that into another app so I could put it in serif font. I was sure it was going to say there's no ethical use of AL.
And I was gonna be like "Yeah, leave Al alone!"
But it says "ai" apparently, and that's a little more complicated. I don't even know what it's supposed to mean. I'm pretty sure there are ethical uses of machine learning algorithms... and probably even for generatice large language models, especially if they can be trained on open uncopyrighted data and used for, like, captioning, and can run on local hardware.
But is that even AI? Or does that abbreviation only apply to the products of Google, Meta, Anthropic, and Open AI?
@Gargron I share your concern, and I’m especially troubled by the use of gas generators in data centers. But I also use devices made with rare minerals, probably ware clothes made by underpaid workers, etc.
I don't think "AI will cure cancer" justifies the harm. A few weeks ago I read (on Mastodon): "Drunk driving is bad and might kill someone, but it often gets people to work on time, so there's a balance…"
Ethics are messy (and I know that’s no excuse).
@Gargron Not even run locally?
@Gargron how about providing **at last** a decent human/computer interface for blind and physically disabled people ? 🤔
@Gargron even if i run #AI local on my 10 years old solar powered #Raspberrypi for translating this toot in every language on the globe and spam it daily into every horrible Tec Bro Prison?
@Gargron Probably right, because it would be a waste of precious resources ;D
❤️
@Gargron I just wonder... do you have children?
Thank you for your answer.