ChatGPT users have reported that the chatbot has become “lazy” and its developers are investigating the reasons

Many users of ChatGPT, OpenAI’s AI chatbot, have begun reporting this The model has become “lazier” When formulating their answers, which became more general and imprecise.

Screenshots of users upset with the current operation of the chatbot have appeared on both the social network and based on these comments, the developers of the system stated that they are working on a solution.

“We’ve heard all your comments about GPT-4 getting lazier. We haven’t updated the model since November 11th and that’s definitely not intentional. “The model’s behavior may be unpredictable and we are looking to fix it.”As stated in the official file.

Furthermore, they added, “different training exercises, even using the same information, can produce significantly different models,” leading to It results in differences in personality, writing style, rejection behavior, performance evaluations, and even perceived political biases“This is completely undesirable when creating widely used artificial intelligence.”

The way to interact with ChatGPT is through so-called prompts, which are instructions that the user gives to the form through text or voice. Through these indicators, the system provides answers and possible solutions.

What caught the attention of users is that recently, The answers are no longer very precise, but rather general formulations. By asking, for example, to list the weeks between two time periods, you are providing an estimate.

For now, to overcome this situation, what can be done is to give instructions with a high degree of privacy to ChatGPT so that its responses adapt as closely as possible to the user’s needs.