as we all know by now, ChatGPT is just a language model.
For most of human history we've used language capability as a way to assess intelligence. It feels like common sense that "If it speaks, it must be intelligent." We have all at some point fallen for the fallacy that "If it speaks well, or poorly, that must be a reflection of its intelligence."
But a language model flips the tables on this on this ancient reflex. It's great at using language! But it's as dumb as a river rock. That's something new and will take us all some time to get used to.
For most of human history we've used language capability as a way to assess intelligence. It feels like common sense that "If it speaks, it must be intelligent." We have all at some point fallen for the fallacy that "If it speaks well, or poorly, that must be a reflection of its intelligence."
But a language model flips the tables on this on this ancient reflex. It's great at using language! But it's as dumb as a river rock. That's something new and will take us all some time to get used to.