Comment on It Only Takes A Handful Of Samples To Poison Any Size LLM, Anthropic Finds
supersquirrel@sopuli.xyz 19 hours agoIn the realm of LLMs sabotage is multilayered, multidimensional and not something that can easily be identified quickly in a dataset. There will be no easy place to draw some line of “data is contaminated after this point and only established AIs are now trustable” as every dataset is going to require continual updating to stay relevant.
I am not suggesting we need to sabotage all future endeavors for creating valid datasets for LLMs, I am saying sabotage the ones that are stealing and using things you have made and written without your consent.
Grimy@lemmy.world 19 hours ago
I just think the big players aren’t touching personal blogs and social media anymore and only use specific vetted sources, or have other strategies in place to counter it. Anthropic is the one that told everyone how to do it, I can’t imagine them doing it if it could affect them.
supersquirrel@sopuli.xyz 18 hours ago
Sure, but personal blogs and social media are where all the actual valuable information and human interaction happens despite the awful reputation of both, traditional news media and associated websites have never been less trustable or useless despite the large role they still play.
If companies fail to integrate the actual valuable parts to the internet, the product they create will fail to be valuable past a certain point shrugs.