3 ms·
Models are not paranoid or anxious, they do not think or have feelings. I know you're probably using those words as a metaphor but we need to be careful about a
by driverdan 6mo ago
Models are not paranoid or anxious, they do not think or have feelings. I know you're probably using those words as a metaphor but we need to be careful about anthropomorphizing LLMs.
- Gareth321 6mo agoAs an accelerationist and transhumanist, no way! These models passed the Turing test years ago. When a thing is indistinguishable from human, it is human. Our brains are, after all, just a collection of learned memetic weights. Just ask the determinists.
- fourside 6mo agoExcept there are several obvious ways in which LLMs are not indistinguishable from humans.
- adammarples 6mo agoThey didn't describe the model, they described (accurately) the behaviour. They are useful descriptors of behaviour.
- selfhoster11 6mo agoThey are trained on natural language. Not anthropomorphizing them is the worse end of the spectrum.