- Openi focuses on what has been wrong in his psychophantic update
- Openi is improving basic training techniques to avoid psychophony in the future
- The head of her model behavior suggests that many personalities in Chat GPT may move forward
Openai has focused on the recent updated rollback to Chat GPT4O because of user complaints about his disturbing, psychotherapy. One in 29 April Blog Post Openi states:
“We have returned last week’s GPT -4O update to Chat GPT so people are now using the previous version with a more balanced behavior. The update we have removed was excessively flatter or pleased – which is often described as a psychinic.
We are actively examining new reforms to solve this problem. We are reviewing how we collect and add feedback of long -term user satisfaction and add them and we are introducing more personal features, which gives users more control over chat GPT behavior.
The blog post is for more reasons why the problem arises, but the height of the issue is that Openi focused too much on short -term feedback, and did not fully calculate how consumers’ interaction with Chat GPT is ready over time.
The Open AI promises to improve its basic training techniques and systems that it indicates the future model to clearly remove the model, as well as allow more user testing before the deployment of models.
Ask me something
By reading the blog post, you realize that the openness feels as if he has dropped the ball on it and is trying his best to edit it.
In addition to issuing a statement, Joan Jung, the head of his model behavior, has just done a job ‘Ask me something’ on Reddate (AMA)Especially about Chat GPT’s personality, psychofancy, and future model behaviors, people’s answers to more questions. The answers revealed interesting insights what the future could be for the Chat GPT personality:
Redded user responsible_ko 2236 asked, “Where do you see the future of the model of the model? Are we moving towards more and more customized personalities, such as giving users the tools tools to create chat GPT sounds and how to communicate with them over time?”
Jung replied by offering the possibility that Chat GPT could come with numerous preset personalities:
“We are working on two things.
The idea of ​​personalities presented in Chat GPT is an interesting and something I can see well with different sounds that have already been presented by Chat GPT Advanced Voice Mode.
As far as we can expect the next update in the Chatgpt-4O, the Open is not saying anything at this time “We are working short and working to correct it.”