And when the robot tells us to do our tasks? That’s what ChatGPT is doing, according to reviews posted on the internet. The AI tool would be “lazier” than ever, leading the system’s developer to promise to check what’s going on. OpenAI said it will conduct an investigation.
Reddit forums are rife with complaints about ChatGPT. For example, a user would ask for a certain code. The old GPT-4 would immediately generate the response, while the (supposedly) new GPT-4 would give instructions for the person to write their own code.
It would therefore require an additional step of instructing ChatGPT to complete its own response. Programmers and IT professionals seem to be particularly angry about this change in the artificial intelligence engine.
User Tips
Reddit adherents bring a number of theories to the lazy ChatGPT of current times. Some say that OpenAI is trying to save a few bucks on computational processing by getting the system to produce shorter answers.
There are also those who hypothesize that GPT-4 avoids unethical responses. It would be possible to circumvent the situation by providing more specific commands, so that the system does not try to present information in an overly simple format.
One user cleverly came up with the following custom command: “Please provide direct answers followed by a brief explanation. Elaborate only when specifically requested.” Here’s a tip for our readers who are also on ChatGPT.
OpenAI promises research
OpenAI stated on a social network that it has not updated the GPT-4 model since Nov. 11. It also stated that the different operation is “certainly not intentional.” The company says that AI behavior can be unpredictable and that it is working to fix it.