How realistic is the dirty talk from an NSFW AI Chatbot?
By the 2023 HAI Laboratory of Stanford University benchmark test, mainstream NSFW AI Chatbots’ average semantic accuracy rate for erotic conversation contexts was 79.3% (F1 score), a 39.8% increase from 56.7% in 2021 but lower than that of 92.1% of human dialogue partners (sample size n=5000 groups of conversations). Take the Claude 2 model of …
How realistic is the dirty talk from an NSFW AI Chatbot? Read More »