From a technical point of view? It’s just software. If you give a software with semi-random outputs access to control external tools, it will do semi-random stuff and could cause harm. And since the people that loves these little runner also likes to run hundreds of them concurrently, with the ability to spawn more, they can cause harm real fast with no oversight.
If you don’t give them uncontrolled access to anything though? Not much harm in that. But it’s harder to sell the “this is the future of everything” if you put pesky things like checks and limited capabilities for the sake of safety.
I almost believe it’s related. It sounds like the left are up in arms over AI not because of safety but because of money. If they would be more specific about safety, maybe people would listen.
safety is a red herring. capitalism is always going to eschew safety in the name of profit, no matter the industry. there is no difference here with ai. just assume the corps to be unsafe with it.
the problem is actually that the corps want the world to “slow down” so they can catch up; “for safety” is just marketing.
I hear a lot about AI safety. What, precisely, is unsafe about AI that is unique to AI?
From a technical point of view? It’s just software. If you give a software with semi-random outputs access to control external tools, it will do semi-random stuff and could cause harm. And since the people that loves these little runner also likes to run hundreds of them concurrently, with the ability to spawn more, they can cause harm real fast with no oversight.
If you don’t give them uncontrolled access to anything though? Not much harm in that. But it’s harder to sell the “this is the future of everything” if you put pesky things like checks and limited capabilities for the sake of safety.
So, nothing unique to AI, except the scale of it.
shareholders losing value when another country they did not invest in taking over. you know that “market and invisible hand”
I almost believe it’s related. It sounds like the left are up in arms over AI not because of safety but because of money. If they would be more specific about safety, maybe people would listen.
safety is a red herring. capitalism is always going to eschew safety in the name of profit, no matter the industry. there is no difference here with ai. just assume the corps to be unsafe with it.
the problem is actually that the corps want the world to “slow down” so they can catch up; “for safety” is just marketing.