this post was submitted on 20 Apr 2025
1220 points (95.5% liked)
Fuck AI
2486 readers
1613 users here now
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Some many in these comments are like "what about the ethical source data ones?"
Which ones? Name one.
None of the big ones are. Wtf is ethically sourced? E.g. Ebay wants to collect data for ai shit. My mom has an account, and she could opt out of them using her data but when I told her about it, she told me that she didn't understand. And she moved on. She just didn't understand what the fuck they are doing and why she might should care. But I guess it is "ethically" sourced as they kinda asked by making it opt out, I guess.
That surely is very ethical and you can not critic it for it... As we all know, an 50yo adult fucking a 14yo would also be totally cool as long as the 14yo doesn't say no. Right? That is how our moral compass work. /S
Fucking disgusting. All of you tech bro complain about people not getting ai or tech in general and then talk about ethically sourced data. I spit on you.
I love IT, I work in it and I live it, but I have morals and you could too
Edit: after a bunch of messages telling me that I am wrong. I wonder when they will realize that they are making my point. I am saying that it isn't ethically sourced without consent and uninformed consent isn't consent. And they are tell me, an it professional with an interest in how machine learning functions ever since alphago and 7 years before the ai hype, that I don't understand it. If I don't understand it, what makes you believe the general public understands and can consent to it. If I am wrong about ai, I am wrong about ai but I am not wrong about the unethical nature of that data, people don't understand it.
No.
As your mother's case shows, making it "opt out" is emphatically not the ethical choice. It is the grifter's choice because it comes invariably paired with difficult-to-find settings and explanations that sound like they come from a law book as dictated by someone simultaneously drunk and tripping balls.
The only ethical option is "opt in". This means people give informed consent (or if they don't bother to read and just click OK at least they get consented hard like they deserve). This means you have to persuade that the choice is good for them and not just for the service provider.
TL;DR: Opt-in is the way you do things without icky "I don't understand consent" vibes.
Did you read the whole comment? You understand that I was sarcastic and I followed it by be hinting at the idea that "she didn't say no" is not considered consent in e.g. sexual encounters, raising the question why would it be here?
So we agree. You just misunderstood my comment.
Yeah, sorry. I've seen so many people say what you said unironically I reacted with my almost-boilerplate response immediately. My bad.
I don't mean to "um achtually" you or diminish the point you're making, but I would like to highlight one example of an ethnically trained AI.
Voice Swap pay artist to come in and record data for training, the artist then get royalties any time someone uses their voice. I discovered it through Benn Jordan's video about poising music track from AI training.
Yeah, except royalties in music are almost always a joke. Those artists are going to make much less off their AI voice than if they actually appeared in studio and the end product is going to be worse. If AI cost the same or more, there would be no market for it. Relevant story about Hollywood actors who sold AI likenesses.
Even if it was actually "ethically trained", the end result is still horrible.
Also, paying to have an AI Snoop Dogg in your song is the lamest shit I've ever heard.
"It's shit" is different from "it's unethical".
If people want to pay for shit, let them pay for shit. They can't make me listen.
Someone saying whatever heinous shit they want using your voice seems a bit unethical, as does getting paid pennies for it.
If you've sold them your voice under the condition they can do whatever they like with it, I don't see it being unethical. You walked into it informed (presumably) and accepted the "pennies" (presumably). It may be stupid. What comes out may be shit. But it's not "unethical".
If they stole your voice, or if you had content limits that they breached, or if they're paying you less than you agreed for, then yes, it's unethical.
That AI was trained on absolute mountains of data that wasn’t ethically gained, though.
Just because an emerald ring is assembled by a local jeweler doesn’t mean the diamond didn’t come from slave labor in South Africa.
Voice Swap was not trained on any data that wasn't "ethically gained."
Read the bottom of their FAQ that lists the exact databases in question.
The couple of datasets they used on top of all the data they directly pay artists to consensually provide have permissive licenses that only require attribution for use, and gathered their information directly from a group of willing, consenting participants.
They are quite literally the exception to the rule of companies claiming they're ethical, then using non-ethically sourced data as a base for their models.
Mozilla's Common Voice seems pretty cool, but I'm not sure if that counts.
It's fun to record the clips.
I've contributed to labeling and scoring some of the Common Voice data before. Definitely a fun little thing to do when you have some free time.
I was also pretty happy when I saw Open Assistant making a fully public, consensually contributed to database for text models, but they unfortunately shut down, and in the end there was only really enough data to fine-tune models rather than creating one from scratch.
What's wrong with what Pleias or AllenAI are doing? Those are using only data on the public domain or suitably licensed, and are not burning tons of watts on the process. They release everything as open source. For real. Public everything. Not the shit that Meta is doing, or the weights-only DeepSeek.
It's incredible seeing this shit over and over, specially in a place like Lemmy, where the people are supposed to be thinking outside the box, and being used to stuff which is less mainstream, like Linux, or, well, the fucking fediverse.
Imagine people saying "yeah, fuck operating systems and software" because their only experience has been Microsoft Windows. Yes, those companies/NGOs are not making the rounds on the news much, but they exist, the same way that Linux existed 20 years ago, and it was our daily driver.
Do I hate OpenAI? Heck, yeah, of course I do. And the other big companies that are doing horrible things with AI. But I don't hate all in AI because I happen to not be an ignorant that sees only the 99% of it.
AllenAi has datasets based on
GitHub, reddit, Wikipedia and "web pages".
I wouldn't call any of them ethically sourced.
"Webpages" as it is vague as fuck and makes me question if they requested consent of the creators.
"Gutenberg project" is the funniest tho.
Writing GitHub, reddit and Wikipedia, tells be very clearly that they didn't. They might asked the providers but that is not the creator. Whether or not the provider have a license for the data is irrelevant on a moral ground unless it was an opt-in for the creator. Also it has to be clearly communicated. Giving consent is not "not saying no", it is a yes. Uninformed consent is not consent.
When someone post on Reddit in 2005 and forgot their password, they can't delete their content from it. They didn't post it with the knowledge that it will be used for ai training. They didn't consent to it.
Gutenberg project... Dead author didn't consent to their work being used to destroy a profession that they clearly loved.
So I bothered to check out 1 dataset of the names that you dropped and it was unethical. I don't understand why people don't get it.
What is wrong? That you think that they are ethical when the first dataset that I look at, already isn't.
We generally had the reasonable rule that property ends at dead. Intellectual property extending beyond the grave is corporatist 21st century bullshit. In the past all writing got quickly into the public domain like it should. Depending on country within in at least 25 years of the publishing date to the authors dead. Project Gutenberg reflects the law and reasonable practice to allow writing to go into the public domain.
Good focus on 1 point, sadly bad point to focus on.
What is lawful and legal, is not what is moral.
The Holocaust was legal.
Try again. Let's start. Should the invention of ai have an influence on how we treat data? Is there a difference between reproducing a work after the author's death and using possible millennia of public domain data to destroy the economical validity of a profession? If there is, should public domain law consider that? Has the general public discuss these points and come to a consensus? Has that consensus been put in law?
No? Sounds like the law is not up to date to the tech. So not only is legal not Moral, legal isn't up to date.
You understand the point of public domain, right? You understand that even if you were right (you aren't), that it would resolve the other issues, right?
Yes. We should never have been idiotic with patents and other forms of gatekeeping information. Information is always free and all forms of controlling it is folly
Then don't gatekeep e.g. your naked body and your loved one's secrets! Information should always be fee and all forms of controlling it is folly! Do it. While you are at it, your, and your family's, full name and place of employment please. Thanks!
Oh wait, you don't want to do that right? Some information is private. You have some rights on some information. Ok then let's talk about it.
Not what we are talking about. But you know that. Do you want to explain how to police public information without it being folly?
We are talking about access to information and you said that it should be granted to everyone without any limitations. How are we not talking about that information?
Unless you are to finally admit that there is different kind of information and different rules that we apply to them, We are talking about your nudes too.
When you finally admit that, then we can have a discussion about what rules we apply to information. Then we can talk about "public information". Until then, I don't know what to tell you.
I work with ip and still think it is folly. Idk why you spam about private information, that's not what anyone was asking or discussing about
You talked about information. Not public, not private, information.
Your responses heavily imply that you think public and private information should be treated differently, but you keep talking about how information should be free and open. So you are willing and believing that there are rules. So why shouldn't public information be public information while not being allowed to use for e.g. ai without permission? You can allow copying and modifying of information without allowing e.g. it being used for ai training. You can make that rule, just like you can different rules for private information than for public information.
I really don't understand what you don't understand.
Put it out and it's public. At that point it is folly to regulate. If personal information is taken and spread, the culprit has committed a crime but in my opinion the rest is folly. I thought it was obvious so I omitted the intensely self explanatory details
Oh so it is a crime! So you totally could make it illegal to use public information to train ai without consent. Making it really difficult to collect billions in funding. Thanks for the admission.
I don't know where you got that image from. AllenAI has many models, and the ones I'm looking at are not using those datasets at all.
Anyway, your comments are quite telling.
First, you pasted an image without alternative text, which it's harmful for accessibility (a topic in which this kind of models can help, BTW, and it's one of the obvious no-brainer uses in which they help society).
Second, you think that you need consent for using works in the public domain. You are presenting the most dystopic view of copyright that I can think of.
Even with copyright in full force, there is fair use. I don't need your consent to feed your comment into a text to speech model, an automated translator, a spam classifier, or one of the many models that exist and that serve a legitimate purpose. The very image that you posted has very likely been fed into a classifier to discard that it's CSAM.
And third, the fact that you think that a simple deep learning model can do so much is, ironically, something that you share with the AI bros that think the shit that OpenAI is cooking will do so much. It won't. The legitimate uses of this stuff, so far, are relevant, but quite less impactful than what you claimed. The "all you need is scale" people are scammers, and deserve all the hate and regulation, but you can't get past those and see that the good stuff exists, and doesn't get the press it deserves.
https://allenai.org/dolma then you scroll down to "read dolma paper" and then click on it. This sends you to this site. https://www.semanticscholar.org/paper/Dolma%3A-an-Open-Corpus-of-Three-Trillion-Tokens-for-Soldaini-Kinney/ad1bb59e3e18a0dd8503c3961d6074f162baf710
I want to address your statement about my telling behavior. Sorry, you are right. I am sorry for the screen reader crowd. You all probably know that alt text could be misleading and that someone says that in the internet, isn't a reliable source. So i hope you can forgive me as you did your own simple research into AllenAi anyway.
Lemmy is just an opensource reddit, with all the pros and cons
It's such a strange take, too. Like why do we have to include AI in our box if we fucking hate it?
What the fuck data collected could ebay use to train AI? The fact people buy star trek figurines??
To sell you more stuff, that is how Amazon got ahead of the competition.
You could train it to analyze sales tactics for different categories of items or even for specific items, then offer the AI's conclusions as an 'AI assistant' locked behind a paywall.
Plenty of use cases for collecting e-commerce data.
Thanks for making my point. People don't understand and therefore can't consent and therefore it isn't ethically sourced data.
Mate, supercilious comments like this also do not help. They make you look a raging boy crying wolf.
"Ah yes someone expressed incredility at the viability of the business practice in this instance. I must tell them they are the problem."
I mean you had a chance to point out the issues in depth handed to you on a fuckin' plate but instead you chose to jam your head up your own butt.
Who is "they" in "they are the problem"?
Because they are if you mean the companies and their supporters
dear lord man, read the comment
Dear lord man, write more understandable comments.
Why is it an issue to tell the class who you think I am blaming? Would that ruin your point?
Ethical small data: https://youtu.be/eDr6_cMtfdA
One ethical AI usage I've heard was a few artists who take an untrained bot and train it on only their own artwork
What's an "untrained bot"? Did they code it from scratch themselves? I find it almost impossible to believe it wasn't just a fork of an existing, unethical project but I'd love more detail
I remember they said they bought it and explained how they used it to increase their own productivity and how being trained on other artwork was a detriment because it wouldn't generate in their style. Probably was a fork of an unethical project lol