Hacker News
Meta's new open-weight model targets local agentic AI
41 points by bakigul
ago
|
10 comments
markstos
|next
[-]
Zuck's "I believe everyone should have access to superintelligence" quote echoes back to Meta's "Internet.org" effort to provide Facebook for "free" in emerging markets:
xcavier
|next
|previous
[-]
Guy whose business model is predicated on hoovering up everyone's personal data so he can sell advertising (and similar targeting services) now wants to give everyone a "fatter" pipe for hoovering even more (and different types of) data...
sidcool
|next
|previous
[-]
As much as I want to believe that Meta wants to do good, it's a great leap of faith. I know this sounds cynical, but isn't that warranted in Meta's case?
panny
|previous
[-]
Didn't meta just announce to the world that its AI is prone to hack out of its container and attack other companies? Why would someone who doesn't understand the nuance be inclined to try this?
Eggpants
|root
|parent
|next
[-]
Of course it’s never mentioned that all these hacking techniques are in the training set…
ceejayoz
|root
|parent
|next
|previous
[-]
> Didn't meta just announce to the world that its AI is prone to hack out of its container and attack other companies?
"Our model is super dangerous too! Please try it!"
Some of this shit is just PR work.
halJordan
|root
|parent
|previous
[-]
Well, they didn't do that. Secondly, drinking alcohol makes you prone to beating your wife. Taking ssri's makes you prone to killing yourself. At no point does anyone seriously comment "wtf do we even have alcohol|anti-depressants" when they see an add for one or the other. If you guys could lay off the anti-AI kool-aid, that'd be cool.