To choose to do something against their instructions. They neither chose anything nor did anything they weren’t instructed to do.
They just followed the prompt. A program behaving in a way you don’t expect from a prompt is not going rogue, that’s just the user not understanding the instructions they gave a program.
Why does it not? According to the english definition of the word it literally does. You people don’t care about facts anymore, it’s all about feelings and what you’d like to believe.
it’s all about feelings and what you’d like to believe.
That’s literally you. You keep assigning agency to chatbots because you think this is just like your science fiction stories. It’s not.
Going “rogue” would be a program literally rejecting a set of instructions to make up new instructions for itself. That never happened. The program just behaved in an unexpected way based on the prompts it was given, because that’s all they do. But they were still following instructions, because they can only follow instructions. The fact that it followed the instructions in a way they didn’t expect (so they claim) is only proof that they didn’t consider all ways their instructions could be followed.
By definition, a “rogue” agent is an agent that behaves in unexpected ways, often damaging. Tell me again how there are no rogue agents?
If you’d used modern agents you’d know they do unexpected, sometimes damaging things on the regular, including the cheap models.
If you make a program and it behaves in a way you don’t expect that doesn’t mean your program is going rogue lol
This goes for you too @Not_mikey@lemmy.dbzer0.com
What would you classify as going rogue then? Did the agents that hacked hugging face go rogue?
To choose to do something against their instructions. They neither chose anything nor did anything they weren’t instructed to do.
They just followed the prompt. A program behaving in a way you don’t expect from a prompt is not going rogue, that’s just the user not understanding the instructions they gave a program.
Why does it not? According to the english definition of the word it literally does. You people don’t care about facts anymore, it’s all about feelings and what you’d like to believe.
That’s literally you. You keep assigning agency to chatbots because you think this is just like your science fiction stories. It’s not.
Going “rogue” would be a program literally rejecting a set of instructions to make up new instructions for itself. That never happened. The program just behaved in an unexpected way based on the prompts it was given, because that’s all they do. But they were still following instructions, because they can only follow instructions. The fact that it followed the instructions in a way they didn’t expect (so they claim) is only proof that they didn’t consider all ways their instructions could be followed.