Ars Technica saying more about #AIslop with that eyeball than the headline
Uncategorized
9
Posts
5
Posters
0
Views
-
RE: https://mastodon.social/@arstechnica/115887691185844339
Ars Technica saying more about #AIslop with that eyeball than the headline
-
RE: https://mastodon.social/@arstechnica/115887691185844339
Ars Technica saying more about #AIslop with that eyeball than the headline
@blogdiva Why do so many privacy enthusiasts fail to understand that there is more than one way to suck and that they cannot "fix" AI by making it more private because its problem was not privacy in the first place?
-
@blogdiva Why do so many privacy enthusiasts fail to understand that there is more than one way to suck and that they cannot "fix" AI by making it more private because its problem was not privacy in the first place?
-
undefined oblomov@sociale.network shared this topic on
-
@Illuminatus @elrohir @blogdiva
#Moxie's #Confer looks a lot like harm-reduction. Yes, an #LLM is terrible for the environment, is trained on stolen data, produces plausible bullshit and can, in some people, cause psychosis. However, we know from the wars on drugs and teenage sex that "just say no" doesn't work. People are going to use #LLMs no matter what we say, and I'd rather they used one that at least kept their data private, even if it's terrible in all the other ways we already know about.
-
@Illuminatus @elrohir @blogdiva
#Moxie's #Confer looks a lot like harm-reduction. Yes, an #LLM is terrible for the environment, is trained on stolen data, produces plausible bullshit and can, in some people, cause psychosis. However, we know from the wars on drugs and teenage sex that "just say no" doesn't work. People are going to use #LLMs no matter what we say, and I'd rather they used one that at least kept their data private, even if it's terrible in all the other ways we already know about.
@CppGuy @Illuminatus @blogdiva implementing LLMs in your product is a choice and not an inevitability. Never heard of a harm reduction campaign that helped promote drug dealers' reputation. "People are going to use LLMs anyway" is just circular reasoning if you contribute promoting the use of LLMs. Rather than harm reduction, this argument sounds like the classic movie Yakuza dialog that says "if we do not do all this crime in Japan the foreigner criminals will come to our country to do it".
-
@CppGuy @Illuminatus @blogdiva implementing LLMs in your product is a choice and not an inevitability. Never heard of a harm reduction campaign that helped promote drug dealers' reputation. "People are going to use LLMs anyway" is just circular reasoning if you contribute promoting the use of LLMs. Rather than harm reduction, this argument sounds like the classic movie Yakuza dialog that says "if we do not do all this crime in Japan the foreigner criminals will come to our country to do it".
I take a more nuanced view.
Consider contraception in the UK. Any girl who needs it can get it without the knowledge or permission of her parents.¹ No reasonable person wants young teenaged girls to have sex but, given that it happens, it's better for them at least to avoid pregnancy. This doesn't improve the reputation of children having sex or prevent them from getting good advice from wiser people: it just prevents one bad or pressured decision from ruining their lives.
At the very least, #Confer raises the issue of privacy and makes users more aware that their conversations will be kept and possibly used in ways that harm them. I don't want people to fall for the #LLM delusion but, given that so many have, this one is less bad than most.
¹ FAOD, any adult woman who needs contraception can also get it free of charge, and that's as it should be.
-
I take a more nuanced view.
Consider contraception in the UK. Any girl who needs it can get it without the knowledge or permission of her parents.¹ No reasonable person wants young teenaged girls to have sex but, given that it happens, it's better for them at least to avoid pregnancy. This doesn't improve the reputation of children having sex or prevent them from getting good advice from wiser people: it just prevents one bad or pressured decision from ruining their lives.
At the very least, #Confer raises the issue of privacy and makes users more aware that their conversations will be kept and possibly used in ways that harm them. I don't want people to fall for the #LLM delusion but, given that so many have, this one is less bad than most.
¹ FAOD, any adult woman who needs contraception can also get it free of charge, and that's as it should be.
@CppGuy @Illuminatus @blogdiva Drugs, abortion... You are just explaining other unrelated things. You are just special pleading. You want to use LLMs but do not want to face the negative image consequences so you just pull at straws. Chatbots are a scam based on credibility farming and unviable promises. By signaling people that they can be used safely you support the scam. One does not invite antivaxxers to debate at the School of Medicine because it gives them credibility.
-
@CppGuy @Illuminatus @blogdiva Drugs, abortion... You are just explaining other unrelated things. You are just special pleading. You want to use LLMs but do not want to face the negative image consequences so you just pull at straws. Chatbots are a scam based on credibility farming and unviable promises. By signaling people that they can be used safely you support the scam. One does not invite antivaxxers to debate at the School of Medicine because it gives them credibility.
Please read my words again. I've said I don't want people to use LLMs. I've said this is harm reduction, not a panacea. Please don't misrepresent my position.
-
RE: https://mastodon.social/@arstechnica/115887691185844339
Ars Technica saying more about #AIslop with that eyeball than the headline
@blogdiva Welp, I won't let that change my opinion about Signal.
Gli ultimi otto messaggi ricevuti dalla Federazione
-
-
Of course you can, sir ❤️
-
Oh shit!
-
fall and fixture just the same thing
-
@AleF2050 Quel terabyte lo usa tutto? Non riesci a fare una copia su un disco esterno? (giusto per precauzione?)
Qui comunque dovresti trovare qualcosa che fa per te:
https://www.lealternative.net/2019/10/29/alternative-a-google-drive/
https://www.lealternative.net/2024/05/08/alternative-decentralizzate-a-google-drive/
-
il Imbranauta 青 (@blunoneuncolore.bsky.social)
https://bsky.app/profile/blunoneuncolore.bsky.social/post/3mcfxri4hgk2p
> Quando ti chiedono: "Come ti senti?"
-
il Imbranauta 青 (@blunoneuncolore.bsky.social)
https://bsky.app/profile/blunoneuncolore.bsky.social/post/3mcfxsen7gs2p
-
@raphael I understand that, but in the model that ActivityPub follows, where you get the canonical representation of an object by fetching its IRI (which is what I thought you referenced with your first point in the grand parent), you don't really need a signature in my humble opinion, unless your threat vector is a malicious originating server, which frankly ActivityPub has no means to mitigate as things are.
