4 ms·
> My reply is that, if it acts as if it understands Chinese in every situation, then there’s no measurable way of distinguishing it from understanding. I'm not
by missingrib 3y ago
> My reply is that, if it acts as if it understands Chinese in every situation, then there’s no measurable way of distinguishing it from understanding.
I'm not sure if you understood the argument. The argument isn't asserting that there is a measurable way of distinguishing it, it's actually claiming that regardless of how well it seems like it understands Chinese, it doesn't actually understand Chinese. It's about intentionality and consciousness.
- int_19h 3y agoThe argument is that the man inside the room doesn't understand Chinese. The entire system still obviously and demonstrably does. In a chatbot, the man inside the room is the LLM, but the whole system is not just the LLM - it's the whole setup that picks generated tokens and feeds them back into the input as a loop. And it demonstrably understands what you tell it, because it can carry out instructions, even extremely convoluted ones or using substitute words that are not part of its training set.
- i-use-nixos-btw 3y agoYou are entirely correct: I do not understand the argument - or, rather, the relevance of it in any applied situation. In effect, my argument is that in order for you to require it to understand something, you require it to understand that thing for a reason. If it acts like it understands that thing under all probing, then your requirement is satisfied - the question about whether it truly understands the thing is moot, because it fulfils the requirement.