• 0 Posts
  • 53 Comments
Joined 1Y ago
cake
Cake day: Jun 10, 2023

help-circle
rss

Maybe he will switch even quicker if you punch him or pull his teeth out.


I get that this is expensive. However, it should also work with RAM if you accept slower speeds I guess. The question is of course if it’s still usable then.



Well, there have been several music lawsuits about certain songs and their amount of identity to others. If you were to write something as closely to another author that you are imitating something like trademark mannerisms there may be a case for that.


I think that writing in someone else’s style to an extent that it becomes very obvious is indeed something that raises copyright concerns.


Zuse built the arguably first electronically programmable computer or something like that.


What an incredibly annoying piece of software. I avoid it wherever I can but it’s unfortunately standard where I work.



True but it might be one of the few reasons websites are even optimised for something else than chrome these days.



The LLM is a tool. It’s like granting copyright to a paintbrush.


That’s not what I meant by that. People should have the rights to the products they produce using the tools at their disposal.


I see a lot of Dunning Kruger here as well. The fact is that you can generate novel images/texts/whatever with these tools. They may mostly suck but they’re still novel so they can be copyrighted by whoever used these tools to create them.


Where did I ever say that a stupid AI should get any rights to its own product?


If it was a compression algorithm then it would be insanely efficient and that’d be the big thing about it. The simple fact is that they aren’t able to reproduce their exact training data so no, they aren’t storing it in a highly compressed form.


They are physically unable to just copy paste stuff. The models are tiny compared to the training data, they don’t store it.



The LLMs don’t deserve or have any rights. They’re a tool that people can use. Just like reference material, spellcheckers, asset libraries or whatever else creatives use. As long as they don’t actually violate copyright in the classical sense of just copy pasting stuff the product people generate using them is probably as (un)original as a lot of art out there. And collages can be transformative enough to qualify for copyright.


In reality people learn how to write lyrics because they listen to songs. Nobody writes a song without listening to thousands of them and many human written songs are really similar to each other. Otherwise the music industry wouldn’t be littered with lawsuits. I don’t really see the difference.


Of course they calculate quite well if it is worth the effort to get rid of fake reviews. I definitely think that they are something that Amazon would eradicate right now if they could do so easily. So as the quantity and quality of fake reviews is bound to rise with recent technological developments, the scales might tip into the direction of them having to do more about it. Because offering good deals as apology is not something that they’d be happy to do for more and more people.


Really? I think it’s not helping an online store if it’s review system can’t be trusted.


Yes, but not a really big one since people should learn how to deal with information and trustworthiness of them anyway


This is what people don’t get. Information is always unreliable when not from a trusted source. Just because it’s easier to generate that kind of information now doesn’t mean it’s a new problem.


But how would they threaten him? What’s leverage could they have against someone on the other side of the world who has the means to protect himself by paying security etc.





So the Saudi investor invests in something that he wants to go broke?


I think they meant the community. Same applies to Wikipedia, not like the site would be worth anything without dedicated users.


Switching between Apple Pay and Cash (Germany). Cash just works, no bank involved, simple. Had some problems where the bank fucked something up and my card stopped working so I had to get cash and pay using that for a week. Cash should always exist.


Oh that’s nice. Are many YouTubers cross posting there?


Are those actually hosting videos or just accessing YouTube? Because for the latter, most people still want the algorithm and the interaction/support to the creators they follow


Any dataset sourced from human activity (eg internet text as in Chat GPT) will always contain the current societal bias.


No I’m just saying that this problem is not a new “AI problem™️” but a basic problem with media literacy that merely gained a new aspect.



People should learn how to approach information sources. If that’s not happening AI doesn’t really matter for this discussion.


That may be true but how are they guaranteeing this on a component level? As far as I know, they use off the shelf stuff so they only have assembly wages. And that doesn’t justify the price imo.


I mean, people should at least check if the publisher is reliable for any information source.


  1. It’s 8 apparently
  2. 5 years is as long as most phones that I used last so there would be no point to buy it