Frankly the best defense is probably to just write your own agent if you’re worried about someone injecting an agenda into one. I strongly suspect most agents would have a place to inject your own agenda and priorities so it knows what you want it to do for you.
There is just a lot of speculation here without practical consideration. And I get it, you have to be aware of possible risks to guard against them, but as a practical matter I’d have to see one actually weaponized before worrying overly much about the consequences.
AI is the ultimate paranoia boogeyman. Simultaneously incapable of the simplest tasks and yet capable of mind control. It can’t be both, and in my experience is far closer to the former than the latter.
Insofar as the agents described in the article, I’m not sure where the overlap with Lemmy is.
they are of many types
I’m just trying to follow the train of thought.
Frankly the best defense is probably to just write your own agent if you’re worried about someone injecting an agenda into one. I strongly suspect most agents would have a place to inject your own agenda and priorities so it knows what you want it to do for you.
There is just a lot of speculation here without practical consideration. And I get it, you have to be aware of possible risks to guard against them, but as a practical matter I’d have to see one actually weaponized before worrying overly much about the consequences.
AI is the ultimate paranoia boogeyman. Simultaneously incapable of the simplest tasks and yet capable of mind control. It can’t be both, and in my experience is far closer to the former than the latter.