OpenAI is actively looking into user-reported issues regarding ChatGPT’s performance, particularly its perceived reluctance to fully engage with user queries. Users utilizing the latest GPT-4 based version of ChatGPT have noticed that it sometimes delivers partial responses or seems uninterested in providing complete answers.
For instance, in scenarios where coding assistance is requested, ChatGPT has been known to provide only basic information, leaving users to finish the task, occasionally responding in a manner described as “sassy.”
These performance concerns have sparked discussions on various platforms, including Reddit and OpenAI’s own developer forums. There’s speculation that OpenAI may have deliberately tweaked ChatGPT to enhance efficiency and reduce the strain on computational resources, as AI systems like ChatGPT are resource-intensive, especially when generating detailed responses.
While OpenAI has recognized these user concerns via Twitter, they’ve clarified that there haven’t been any recent modifications to the model that would lead to such behavioral changes. The company is investigating these claims but hasn’t yet confirmed any alterations in ChatGPT’s response patterns.
Sources include: The Independent’s article on ChatGPT’s performance