According to news on December 12, more and more users have recently complained that the latest version of OpenAI’s chatbot ChatGPT has become “lazy”. OpenAI responded that it was investigating the relevant complaints. The latest version of the chat robot ChatGPT is based on OpenAI’s GPT-4 large language model. Many users complain that the current ChatGPT seems unwilling to complete tasks according to their requirements, or seems not very interested in answering their questions.
For example, if a user asks to generate a piece of code, ChatGPT may only provide a little information and then ask the user to fill in other details themselves. Some people jokingly complained that ChatGPT was a particularly snazzy way of telling people they could do the job themselves.
On the social media Reddit and OpenAI developer forums, users have complained that ChatGPT is not as easy to use as before. They also speculate that OpenAI did this on purpose so that ChatGPT would be more efficient and not give too long answers.
AI systems like ChatGPT are expensive for the companies behind them, so giving detailed answers to questions can require a lot of processing power and computing time.
OpenAI, the developer of ChatGPT, said they are aware of complaints about the system. But they said they had not actually made any adjustments to the model and could not explain why ChatGPT performed differently.
OpenAI said on social media
OpenAI did not say whether it acknowledged the phenomenon or whether ChatGPT itself had changed the way it responded to the issue.
Last month, OpenAI CEO Sam Altman was forced out of the company, only to be reinstated days later. The company has been in turmoil in recent weeks.