An important update from OpenAI on the evolution of ChatGPT: we are moving towards a plurality of “cats” with different worldviews (and not the same moral values) https://t.co/GThvFZI6f3
In the new version there will be in addition to the basic conversational model, conversational models adapted to different expectations (more or less moderation and, to put it clearly, more or less right). pic.twitter.com/Z0UI70NYuH
More deeply, large-scale tests of chatGPT and, more recently, of Bing show that it is not going to be easy to improve the reliability of the generated texts. Creating a plurality of cats also means taking refuge behind a certain relativism…
With different shades of ChatGPT, tolerance to hallucinations will probably be higher: it’s closer to a talk show, rather than an encyclopedia, where the speakers do not necessarily verify what they say. For OpenAI this is clearly welcome.
