Raw LLM Responses

Every few months since this video released I get baffled by hearing about or even experiencing advances in ai which a few months ago seemed like far-off sci-fi stuff, that won't happen before a decade even by the most agressive estimate. And turns out now it's just fucking there I don't know if it's the newest, probably not, but a chatbot newer than chatgpt I spoke with really seems human. I tried messing around and testing it, and there's a hard to describe something about the cheery tone it takes, but apart from that if I didn't know it's a chatbot, I would never guess What did trip it up was logic stuff "my father's son's dad has a daughter" and for some reason it can't solve math that chatgpt can, but apart from that there's nothing that it didn't know how to react to. What was unhumanlike was that it had great knowledge of science, didn't mind switching topics erratically, and asking about gravity and freefall gets you a textbook explanation. It also was susceptible to the thing gpt does, that when asked about a piece of fiction if it doesn't know the answer it'll make up something "ah, yes, the ending of 20000 leagues under the sea was indeed very cool, I was in tears when captain nemo reunited with his dad" Not long ago you could just say "this ain't it chief" and show any response of chatbots to show their flaws, now you've gotta describe each subtle quirk that an otherwise human sounding thing has