It’s distasteful, but a useful weapon. We need ways to create content that is good enough and really fast to make. A grassroots movement doesn’t have the funds to pay marketing teams to do our work for us.
AI is being used every social media site the world over to hamper discourse, embroil people, to enrage people…
I don’t know of another way to combat it except with more AI. Maybe at some point people will completely get off of the internet with the thought “There’s no real people on the internet any more”.
Or here’s a discussion on r*ddit about how Glaze is totally ineffective against poisoning image generation AIs. (Same principles apply to others like Nightshade)
Basically every poisoning technique is by its very nature too specific to be useful for poisoning all the types of AI out there, or is too invasive and ruins the content for humans too, or is not strong enough and it gets lost in the cacaphony of data anyway.
Really it’s a clever idea in principle but in practice AI poisoning isn’t really a mathematically viable way to combat AI.
You can’t avoid AI or things that use AI. We’ve been using these statistical models WELL before they became mainstream. And they mimic human authors well enough that it’s impossible in many cases.
It’s distasteful, but a useful weapon. We need ways to create content that is good enough and really fast to make. A grassroots movement doesn’t have the funds to pay marketing teams to do our work for us.
What did people do before AI? It wasn’t that long ago.
AI is being used every social media site the world over to hamper discourse, embroil people, to enrage people…
I don’t know of another way to combat it except with more AI. Maybe at some point people will completely get off of the internet with the thought “There’s no real people on the internet any more”.
You poison it and you dont use it. That’d how you fight AI. With tarpits and poison. They’re like Mentats.
That doesn’t work, based on the results we’ve seen so far. All attempts to poison AI have quickly been bypassed.
Is there any source to this claim?
Sure, a quick google search turned up this, published just four days ago:
https://ieeexplore.ieee.org/document/10962550
Or here’s a discussion on r*ddit about how Glaze is totally ineffective against poisoning image generation AIs. (Same principles apply to others like Nightshade)
https://www.reddit.com/r/aiwars/comments/12f9otc/so_the_whole_entire_glaze_ai_thing_does_it/
Basically every poisoning technique is by its very nature too specific to be useful for poisoning all the types of AI out there, or is too invasive and ruins the content for humans too, or is not strong enough and it gets lost in the cacaphony of data anyway.
Really it’s a clever idea in principle but in practice AI poisoning isn’t really a mathematically viable way to combat AI.
Ban AI where you can and talk about how bad it is when its relevant
They are never going to ban AI. It replaces too many workers. Hell, it’s smarter than half the people I know – even when it makes mistakes.
Who is “they”?
But i know AI wont be banned in most countries ever. The point is to try to make people avoid AI or avoid things that use AI
You can’t avoid AI or things that use AI. We’ve been using these statistical models WELL before they became mainstream. And they mimic human authors well enough that it’s impossible in many cases.
Its impossible in some cases, but possible in others
If people are openly using AI, others will be encouraged to do that to