Openai CEO Sam Altman admitted in a post on X on Monday that recently the chatup -4 O update has made the personality of the AI ​​model a little “psychophent -Y and annoying”. Altman promised to decide soon. On Wednesday, Openai CEO shared a fix: it is rolling back the update. Altman says that updates have been rolled 100 percent for chat users, and will soon be fully rolled back to free users. He says that as soon as the roll back is complete, a new update will be released.

In addition to fixing the “annoying” behavior of the model, Openai will soon start some other updates for GPT-4o’s personality. However, the details will be shared in a few days. Ultman wrote in a post on X, “We started returning the latest updates for GPT -4O last night. Now it is back to 100% free users and when it is finished for users paid, we will update again. Today we are working on additional reforms for model personality and will share more in the coming days,” Altman will write more in the coming days.

Earlier this week, when Sam Altman accepted the new annoying personality of the Chatgpt-4O model, he teased that the AI ​​model could soon get the ability to switch personality. He disclosed this in response to a comment on his post on X, where a user asked if he would soon be able to change the personality of AI Chatbot, or if “old and new (can be personality) are distinguished in any way?” Altman replied, “Yes, we finally need to be able to offer many options clearly”.

When Altman wrote in his post today that “we are working on additional reforms for model personality”, can the option to switch between personality can be indicated? Perhaps. We don’t know. But we will soon learn about it in “coming days”.

Openai rolled out its new Chatgpt-4O model in May 2024, leading to a major performance upgradation on the GPT-4 turbo. The company claimed that it was twice as fast, the use range was five times higher, and the cost was half. What set chat -4o was different, lessons, audio and now images had the ability to work originally -it makes multimodal. It can deal with complex mathematics, explain facial expressions, and translate immediately spoken interactions. With a recent addition of the image generation tool directly on the GPT-4O AI model, users can now interact in all major formats in a more comfortable, human-like manner.