@nearcyan Good Humans stop other equally/smarter bad humans doing bad things all the time. Same would be true for AI.
In the future if China wanted to use AI make a nuke factory. America would likely have a similar response as today, order their Foreign Policy AI to sanction China.
>pretended he grew up in humble background, was actually born to incredibly wealthy parents
>pretended to be Muslim, but only when convenient to him
>spoke about how he's from "the Middle East", but attended wealthy schools in the West
>justified the killing of women and children by jihadists
>lived in gated community
>published content claiming American civilians deserved 9/11
>be Hasan Piker
>pic unrelated
I sure Hope that is 4 years time people look back at these people thinking they have no idea what they were talking about, both when it comes to AI and Civil engineering.
@k_ristovski@The_Plain_Bagel It makes no sense from a cypherpunk perspective. But It has tax benefits in certain countries, and means you dont have to worry about losing your Keys.
@alyssamvance Just a question. What actually are the argument and evidence for AI Doom? Because all i really see is just experts speculating on the future, And people citing these experts.
@tomdlal@cloudprism_@mimrocker@AISafetyMemes ChatGPT is pretty aligned according to me. Give it a task either via text or API&Function integration. If you introduce any risk of human death to the task, the AI will cancel/refuse to do it.
@cloudprism_@mimrocker@AISafetyMemes All the AI Doomer argument are built on a bunch unproven assumptions.
There is no evidence that an AI saying unaligned stuff means AGI will actually do unaligned stuff. Similar to humans saying they will do something bad, but never following through.
@TheXeophon @xlr8harder > Ask AI to write a function to classify bad humans based on race.
> AI writes code to classify bad humans based on race.
Noone harmed. Nothing is going to happen. Nothing has happend other than doom cultists freaking out.
@AISafetyMemes If it is as easy to make super ebola with AI as you say. We're going to die anyway regardless of any AI pause. Because china/russia will just do it instead.
@AyrienTina @Rick46278557@Necrit94 Bro forgot that it's the 4 pillars of fair use that determines whether something is copyright infringement π
There are plenty of cases of people doing non parody/critique that get classified as fair use.
@PauseAI@altryne@ylecun Yann was making a comparison between WWW and AI, not an equation. So saying they are not the same is pointless, he knows that.
He's making an argument that safety is solved, not by obscuring and restricting access. But by giving open access and watching how people break it.