Ope opening Was given, in May last year, the Mato GBD -4 O To understand the audio, image and text in a unique way, provide quick intelligence, understand and analyze shared photos, allow natural conversation in real time, and improve skills in text, voice and vision. This model is widely available to users, however, its latest update presented a failure a Behavior “very praise and annoying” In the experience of use.
Sam Altman was shared in his X profile (formerly Twitter), where “the latest GPT -4 Ou updates are very annoying personality,” he said, however, he added They will work “additional corrections for model personality”.
To adjust this ‘uncomfortable’ experience of the app, Open Openei changed to launch the last update Free users and users have signed the premium version, waiting for corrections to connect the model personality. In addition, in this sense, Altman’s company made clear in A Official report It is, by eliminating the “excessive praise or pleasant update of the GPD -4 Ov”, users They used the previous version with “very uniform behavior”.
On the other hand, opening new solutions to avoid annoying behavior, Openly checked the user’s satisfaction for a long time More customization introduces functions SATGPT must have more control over behavior.
What exactly happened to the Saadget?
In this update, the above report explains the opening He was “paid more attention to short -term ideas” And he does not consider “how the user contacts with the Chadgipt be formed over time.” So, as a result, the “GPT -4 o was very favorable, but small honest” answers.
“Chadjpt’s predetermined personality deeply affects the way you enjoy it and trust him. Flattery’s interactions are uncomfortable, confused and can cause pain. We are not successful and we work to solve it. Our goal is that Saadgift helps users to explore ideas, make decisions or showcase the possibilities, ”says Open Hey.
Sign up with us Newsletter Get the latest messages on technology in your post.
