OpenAI has confirmed a noticeable decline in the efficiency of ChatGPT, primarily attributed to the GPT-4 system. This acknowledgment comes in response to a growing number of user complaints regarding the AI’s performance, which has been perceived as increasingly ‘lazy’ or less responsive.
OpenAI has pinpointed the lack of updates since November 11th as a primary cause of the altered behavior in ChatGPT. The AI’s performance unpredictability, a result of this stagnation, has led to various issues. Users report having to repeat instructions, experiencing delayed responses, and encountering error messages. Developer Rohit Krishnan expressed frustration over the AI’s limited capabilities, highlighting issues like truncated tasks and restricted functionalities.
The change in ChatGPT’s performance has significantly impacted user experience. One user described the situation as akin to driving a high-performance vehicle that suddenly downgrades to an old, inefficient model. OpenAI has requested patience from its users, explaining that training AI models is a complex and nuanced process. Variations in training can lead to noticeable differences in the AI’s personality, affecting aspects like writing style, evaluation performance, and even political bias.
OpenAI has encouraged users to continue reporting any problems they encounter with ChatGPT. The organization is actively working to enhance the model’s capabilities and usability for a wide range of applications. In the meantime, users have been finding creative ways to work around these issues, including using specific instructions to avoid repetition and employing emotional stimuli to boost the AI’s performance.
OpenAI recently launched GPT-4 Turbo, an update to the existing model that includes data on events up to April 2023. This enhanced version is available to subscribers of ChatGPT Plus, offering a more refined user experience.