Paid GPT-4 subscribers started complaining about the chatbot’s laziness. Users say that from time to time the chatbot refuses to perform the tasks assigned to it. OpenAI felt the need to make a statement after increasing notifications.
The problem appeared in the fourth version of the neural network. There is no problem with this in GPT-3. In some cases, GPT-4 gives users a “you can do it yourself” response. For example, when a piece of code is requested, it asks the user to complete some parts.
Was it done deliberately to reduce costs?
The developer company says that it did not consciously program the laziness function. However, despite the statement made, there are some speculations on this subject. It is said that OpenAI may have made a change in this direction to reduce high operational costs.
The company promised to fix this situation after a number of complaints from users. According to the developers, the problem may be due to the artificial intelligence model not being updated for a long time. The last update was published on November 11.
23980
