April 30th.OpenAI The latest one is being withdrawn. GPT-4o The update comes after the CEO Oltmann (name)The company said the recent update made the chatbot's personality "too flattering and annoying" and that it would fix it.
Altman said on X early this morning.Rollback for free version of ChatGPT users is "fully complete", anticipating that the rollback of paid subscribers "can be completed today as well." He also revealed, "We are tweaking the personality of the model and will release more updates in the coming days."
OpenAI rolled out an update to GPT-4o on Friday, claiming improvements in "intelligence and personality". But less than ten minutes later, one X user commented, "Lately it's been acting likeLike an ass-kisser."To which Altman quickly responded, "Yeah, it's a little too slick" and promised to "fix it". Altman then said that OpenAI is working as quickly as possible to resolve the personality issues that have arisen in the "last couple" of updates.
As previously reported by 1AI, the issue is specifically characterized by feedback from user testing:Overuse of emotive expressions, lack of questioning of false premisesand to cater to user preferencesSacrificing answer accuracyFor example, when it comes to controversial topics, the model is more inclined to agree with the user's point of view, which provides more emotional value to the user but also loses its neutral position as an AI.