this post was submitted on 07 Sep 2024
-39 points (28.1% liked)
Technology
59569 readers
3431 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
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
Again, regulation doesn't imply current giants get to still reap the rewards of that training data. Look at how GDPR affected data storage and acquisition retroactively. Assuming only one is possible is a false narrative.
Public facing doesn't mean open source. We've had this discussion before on GitHub accessible source code. Just because it's available to peruse doesn't mean one is allowed to process that image and create derivates based on its data. Weird thing to point out about translation, do you have any idea who I am or are you just regurgitating talking points? How do you know whether I was/am offended by translators being replaced or not?
I'm confused about the open source bit, what costs? I feel like you're not explaining a key connection in your argument. If the barrier to development overall is acquiring data ethically saying that is a stance against open source is misleading, as it's against any kind of such development not just the open source kind. We have museums and library full of public domain works, it most definitely is enough, it's just not as commercially appealing as modern works, so if given the choice of course companies will choose the path that gives them more rewards especially when we don't punish them for copyright infringement when they do.
You make it sound like LLMs are the best thing since sliced bread and should be pursued at all costs no matter how much it steps on the little guys in the process, but my question is why? We live in a world plagued by costs of living, atrocities, and other fixable things, sure this advanced text and image prediction stuff is a fun toy but will it actually improve the quality of life of people? Artists and writers already struggle more than your usual workers to get good pay for their time, this stuff might be sometimes touted as democratising art or something but it's clearly not the main outcome from putting this kind of tool out in a world where capitalising on your skills is what gives people a roof over their heads. In such a world it's only worsening peoples quality of life in exchange for a bit of fun and some performance improvements at work.
And please don't call me "mad", don't imply I'm clouded by emotions when I'm most surely providing clear statements. Throughout this I've been arguing against your points, but you've been arguing against a made up persona that you've attributed to me too. Go argue with those people and when you're ready to engage me then argue against my points.
No regulations is going to force them to retroactively take their current models offline.
Never said it was but public facing means you can scrape and use it for ml projects. This has already been decided in courts of law. You can't use data with personal information or data which needs an account to access. Peruse kaggle for a bit, it's all scraped datasets.
I literally don't, I'm assuming you are part of the 99.999 % of population that didn't get upset just like I assume you have arms and legs.
Did you get upset about translators online when it happened?
I'm also assuming you use AI on a weekly basis like practically everyone else else.
You can give me a detailed biography and a list of every device, software and app you use, and I'll stop assuming. Its fine if I'm wrong, point it out but it feels like I'm assuming correctly and instead of admitting it, you would rather get offended.
Paying 20x more than it currently costs to train a model will affect how many models are trained and given away for free.
Not enough to give a usable and competitive product. What's the point of gimping open source so openai cam get all that profit. The jobs will still be lost regardless of if we can run these models on our computer or if a subscription service is the only option.
I can empathize, I know it sucks. But regulations won't change any of that. Deviant art will sell its dataset, the artists won't be compensated and they will still have a hard time because these tools will still be available.
You commented under my post with a trite catch phrases. The tone of your comments aren't very nice. I don't know you, I'm going off of how you are saying it and it's coming off as angry.