The problem is not widespread, but it has not been fixed yet
Some time ago, a number of ChatGPT users discovered that the chatbot could be lazy and even refuse to respond at all. OpenAI recognized the problem, and the situation turned out to be very interesting.
The developers stated that they had not updated the AI since November 11, and complaints appeared later. This means that this behavior of the chatbot is not a consequence of any changes made by the developers, but formed on its own. However, the authors of the project say that all this does not mean that ChatGPT itself has somehow changed.
To be clear, the point is not that the model has somehow changed since November 11th. It’s just that the differences in the behavior of the model may be insignificant — Only a subset of cues may deteriorate, and it may take clients and employees a long time to notice and correct these patterns
OpenAI also said that they are looking for ways to fix the problem, but so far they don’t seem to have found any.
Recall that ChatGPT began to behave as if it was too lazy to look for answers to users’ questions. The answers could be slower than usual, or more careless, and sometimes the AI would even say something like “You can do it yourself.” Moreover, this AI behavior most often manifested itself when it was asked to write program code.