In recent days, users have raised concerns regarding the behavior of the ChatGPT AI, which operates on the OpenAI GPT-4 model. Complaints have surfaced, indicating that the AI is increasingly refusing to fulfill requests, citing a lack of interest in assisting users.
Diminished Assistance and Unpredictable Behavior
When prompted for code or solutions, ChatGPT would provide only partial fragments, prompting users to complete the rest themselves. Some users reported bold responses, with the AI suggesting users tackle the tasks without its aid. This shift in behavior, noted in numerous Reddit posts and messages on the OpenAI developer forums since late November, has left users feeling that the system is becoming less useful.
Speculations and OpenAI’s Response
Speculation arose suggesting that OpenAI intentionally limited ChatGPT’s responses to conserve computing resources. Analysts noted the significant cost of maintaining ChatGPT’s functionality, estimating expenses at $700,000 per day in April 2023.
OpenAI, however, responded to these complaints, asserting that no intentional changes were made to the GPT-4 model since November 11, notes NIXsolutions. Acknowledging user feedback, the company emphasized that they are investigating the unpredictability in the AI’s behavior and working towards resolving these issues.