Meta admits that it trains its AI on your Instagram and Facebook posts::undefined
Every time I read a new story about another example of this, I struggle to understand why people are getting outraged. Did you have some expectation of privacy when you published your thing to the world-connected network run by profit-seeking corporations?
I don’t understand why anyone cares, nothing I posted on Facebook has any value to me at all so if someone can use it to train an AI then good on them.
Its media creating a new boogie man. LLM and AI gives regular people a lot of crazy power and creative ability that can challenge current power dynamics. So media is making sure regular folk fear the shit out of it.
We’ve reached AI is coming for the children articles. AI is stealing your jobs. AI is going to murder your favorite comedians. AI will turn you into a battery. Everyday its some fucking new fear mongering headline
LLM and AI gives regular people a lot of crazy power and creative ability that can challenge current power dynamics.
My view is opposite in that LLM and AI will further entrench the skewed power dynamics, as only really the big companies like Microsoft, Amazon and Google can fully exploit it, drive and and afford it. Sure I can run some LLMs on my computers at home, but you really need access to a lot of data and computing power to create the models in the first place.
An LLM trained exclusively on Facebook would be hilarious. It’d be like the Monty Python argument skit.
That’s where LLaMA came from and it’s actually pretty good.
It uses other datasets. I think FB is mostly for training on how to deal with emotional stupid people and incorrectness.
I dunno about that, going by the shit my egg sack posts…
So it’s racist and can’t spell
😂
So Meta’s AI is an angry, racist, self-centered, conspiracy theory obsessed, anti-based, flat-earth, quasi illiterate, artificial intelligence? I always wondered how and why Skynet decided to wipe out humanity. Now I know.
deleted by creator
No wonder it’s terrible
So the AI will keep trying to do searches in its status update bar?
Cue the drop table fb posts
I think at this point we can just assume that everything Meta does is the worst possible way that thing could have been done.
I’m no fan of meta, but a reminder that they are one of the best right now for keeping their AI developments more open and available. This is thanks to yann lecun and other researchers pressuring meta to keep their info on the subject more open.
Are we looking to punish them for making their work accessible?
Not to mention how important something like joint embedded predictive architecture could be for the future of alignment and real world training/learning. Maybe go after other foundation model developers to be more open, if we’re complaining about the inevitably public nature of some information within the mountainous datasets being used.
Although I’m still of the mindset that the model intent matters more than the use of openly available data in training. I.E. I’ve been shouting about models being used specifically to predict and manipulate user interactions/habits for the better part of a decade. For your “customized advertisements” and the like.
The general public and media interaction on the topic this past year has been insufferably out of touch.
Isn’t it only open source because of a mistake or a leak? I thought Meta planned on it being 100% proprietary but then tried to spin the leak as them being open.
Not a fan of FB, but this is way overblown AFAIC, you post something publicly expect it to be used publicly
The AI is going to be weirdly emo and phrasing things like “X is not sure he can do it” because my Facebook activity skews heavily for 10ish years ago when it was popular.
