fantasizing about getting rid of AI

[Cipower3]

Im sick of this shit. all these stupid AI promotion's saying how its going to change the world. Like YaEh, Companies being able to mass spread disinformation and piggyback of the work of creative people so its easier to skrew them over is going to fucking SAVE HUMANITY. all of that is just to make the average joe smo comfortable with this. And since it seems that nowtimes people have developed an allergy to thinking about it they make exuses like (but it looks pwitty) or (oh is so much more effisient. ) and i hate this becuse it promotes this idea that everything is just content and that you should just be a brainless consumer that shoulednt put effort in to things and that the only thing that matters in NumBeR BeeG. its just a cycle initiated by greed of the wealthy and perpetuated by the gluttony of the public. so i wanted to just imagine some best case senarios where the people who matter could get their way.

All of this stuff is basically impossible, and one of those tracker people form google have probably allready found this and

making protections as we speak.

if only we could give the Ai servers rats to delete the models and replace them with goreor we send zip bombs containing petabytes of infecten audio and images the multiply themselves. giving the AIs aids. or even craizeir we mess with the learning prosses so that the ais would try to be as offencive as possible to whoever is using it. imagine at some huge press conference when the reveal the newest version of Chap GPT 5 and the thing just says "I. Dislike. Black.... people."

then when they tell it to stop it just does it more and the supper poisin files corupt all the other ones in the model so it just does it more. and before you know it, every ai the caught internalized homofobia and just spews alt right rederic everywhere. news articles, televises, broadcast, PSAs, the whole shebang. all the ais and the comanys reputations are damaged beyond repare and they have to pay us back for all the drawings, music, writing, and everything else the scraped of our websites and post. They will have to rely on us to porvide them their fincances. and in turn the give actual artist the final say in what projects get greenlit and make every social media feed the way the feed on newgrounds works where its basicly non existant. just that last one would make things a whole lot better.

and then we behead Sam Altman and piss on his corpes.(this last one is a joke.)

[detergent1]

what you describe has happened before, see the Tay incident: https://knowyourmeme.com/memes/sites/tay-ai

a virus for the cloud is possible, but cold backups are unlikely to be damaged, so the damage is easily reversible

LLVMs do learn about very bad stuff such as gore, racism and homophobia, they are already feed with terabytes of such content, but the catch is that they are trained with human-in-the-loop feedback: employees and community manually review thousands of conversations, rewarding and punishing the machine until it behaves appropriately, so it is no use to feed it with more bad content

as for protection of artists, writers and other professionals: all we can do is demand regulation and laws from our legislation. for example, Brazil has banned AI dubbing of movies to protect our dubbing industry.

[alsoknownas1]

At 12/6/24 03:57 PM, Cipower3 wrote:Im sick of this shit. [...]

The following is entirely my fault for responding to yet another NG AI thread.

Look, I'm sympathetic to your fear. I regularly put my dick inside an artist and said dick is attached to a programmer (me). The thing is that you need to pull back and see that art, animation, coding, even writing are exactly nothing percent of modern AI's use cases and that it's real uses are myriad and already happening. For that reason, AI is here to stay and is only going to get more ubiquitous. 

Either at my own employer or at offices where friends work, AI is already being used to:

• transcribe all meetings so that they are as easily searched as chats
• eliminate traditional phone trees
• digitise local flood records going back a hundred years
• improve screen readers so that even badly designed software can be used by the vision impaired
• do sentiment analysis to find abusive and threatening messages or to identify individuals that might be suicidal

When normies think of (and pay for) "AI", these are the sort of things they think about and buy. Animating children's cartoons or comic books is the furthest thing from their minds.

I know this was just meant as an emotional vent to what you hoped was a sympathetic audience and fair enough. The thing I would say to those trying to carve out special protections like the ones @detergent1 mentions is that you need to understand that while you hear about AI on the Internet--a content channel--the main uses/market for AI is not in content. Content is a small part of the overall economy. I understand that to the content centred industries AI seems apocalyptic, but please see the recent US election for a concrete example of how such "the sky is falling, this is evil" framing does not actually advance political causes, however good it may feel in the moment.

For better or worse, the only way I see regulatory AI carve-outs for creative jobs happening in the West is under "cultural preservation" a' la the way we fund opera and ballet or the X% local programming protections you have for losing cultures like Australia and Canada.

[AypS]

Well, I've heard some AI are starting to train on AI creation. With those incestuous creation, it is possible that they might do it themselves !

Another possibility of "they might do it themselves" is the fact that most AI company don't work enough to solve the "bias" problem. This problem is the difference between what you ask to an AI and what it understand, and unlike what it seems, we are far from a solution.

About the "getting around AI protection to make it learn shit", you should try this. It might intrested you.

[detergent1]

At 12/7/24 12:58 AM, alsoknownas1 wrote:Either at my own employer or at offices where friends work, AI is already being used to:

[list]

are you using LLMs and Stable Diffusion to solve these problems? because I think OP is worried specifically about generative text and generative images etc. I don't think anyone has beef with more classic AI at all, not even OP

[Redmage4]

Students are using AI right now to do homework, write essays and even even get a degree. Like it or not, our children and grandchildren might actually have to face terminators if things go wrong. So what will happen? This...

https://www.youtube.com/watch?v=icVy7Ve6y6A