OpenAI released ChatGPT, a prototype of a general-purpose chatbot, which can start conversations with users on various topics. ChatGPT is based on OpenAI’s GPT-3.5 model, which can talk fluently like humans, showing the great progress of chatbots in the past few years. It also demonstrates the flaws of chatbots, presenting false or false information as fact. Chatbots are still some sort of “random parrot,” the AI researchers explained, whose knowledge comes from statistical regularities in the training data and which understand the world differently than humans. It can help users debug code, say something wrong with a loop condition in a casual tone, or explain a bubble algorithm like a gangster, it can write sitcom scripts, explain different scientific concepts, write academic papers, Possesses extensive knowledge. But it’s still your typical chatbot, shying away from discussions on sensitive topics.
This article is transferred from: https://www.solidot.org/story?sid=73544
This site is only for collection, and the copyright belongs to the original author.