Openai announced late Friday night that he was updating his latest model to be “warmer and friendly”.
The company recently launched the much awaited GPT-5 in a process, which CEO Sam Altman admitted that “a little more bumpy than what we expected,” some users complained that they prefer the previous model, GPT-4O.
Openai is trying to address some of the complaints with this update, it is with changes that say it is “subtle”, but will make GPT-5 “more acceptable” now.
The company wrote in the social media post, “You do not make small, real touch ‘good questions’ or ‘great start,’ don’t flatter.” “Internal tests show no increase in sycophancy compared to previous GPT-5 personality.”
In a dinner with reporters this week, Openai officials tried to focus on the company’s plans beyond the GPT-5, but according to the Max Zaf’s report, the Rocky launch was an elephant in the launch room. As far as the model friendship is concerned, VP Nick Turley said that GPT-5 was “just on this point”, but that the new update-like-like has now been announced-feel hot.
