Have you ever imagined a situation in which the robot tells you to carry out the tasks it was designed to do? Because this is what users of ChatGPT are reporting on social media in recent days. According to users, artificial intelligence (AI) would be extremely “lazy”, which caused Open AI, the system’s developer, to open an investigation to verify what is happening.
See also: In the ChatGPT queue? 2 other options to use while you wait
Given the situation, users are using various forums on Reddit to talk about the problem and look for solutions. One of the internet users reported that he requested a certain code. While the old GPT-4 immediately generated the requested code, the new GPT-4 gives instructions for the Internet user to write their own code.
Therefore, it would be necessary to add an extra step to induce ChatGPT to deliver the ready-made code, instead of instructions for the user to perform the task. The main complaints come from programmers and IT professionals who use AI tools a lot to optimize their work.
Users give tips to overcome the problem
Reddit users commented on several theories to explain what has been happening with ChatGPT. While some claim that OpenAI is trying to save on computational processing, others hypothesize that GPT-4 is avoiding giving unethical answers.
However, in both cases, users are looking for possible solutions to overcome the situation. One of them is to offer more specific commands, so that the system does not try to present very simple answers. So, one user created the following custom command to try to receive more direct answers: “Please provide direct answers followed by a brief explanation. Elaborate only when specifically requested.”
Finally, OpenAI reported that it has not updated the GPT-4 model since November 11th, justifying that the different behavior of the AI “is certainly not intentional”. Thus, the company stated that the functioning of AI can be unpredictable, but that it is already working to correct the flaws.
2023-12-17 23:24:00
#ChatGPT #lazy #Chatbot #refusing #give #direct #answers
