ChatGPT has become so lazy that it asks you to do the work for it


For several weeks, ChatGPT users have been complaining that the AI ​​has become “lazy”. When asked to carry out relatively difficult tasks, she now simply gives instructions on how to carry out the task herself. OpenAI has acknowledged the issue and indicated that a fix is ​​in development.

keyboard-chatgpt
Credit: 123rf

When ChatGPT was revealed to the general public, the consensus was that the tool would revolutionize our daily lives, by relieving us of boring tasks to save us time. A year and a few later, it seems that AI is tired of doing us favors. As surprising as it may seem, several voices have been raised in recent weeks to denounce the “laziness” of the chatbot.

Indeed, it seems that ChatGPT is now reluctant to accomplish certain tasks, in fact even the most difficult tasks. Instead, the chatbot prefers to give advice to the user on how to carry out their request themselves. On X (Twitter), Matt Wensing shares his conversation with the AI, during which he asked it to list the weeks until May 5, 2024.

On the same subject — Oops! ChatGPT reveals phone numbers and private addresses

ChatGPT also needs a vacation

The interested party’s response: “I cannot generate an accurate list, […] but I can give you a rough estimate.” Not very satisfactory, especially since the task is largely within ChatGPT’s scope. Subsequently, Matt Wensing pointed out the latter to him, following which ChatGPT finally deigned to respond to his request.

On social networks, testimonies of this type have multiplied in recent days. Another user indicates that by asking ChatGPT to generate a code of around fifty lines, the tool generated a code page with, instead of the response, an insert “insert your code here”. Bordering on insolence.

Although there are solutions to ensure that ChatGPT ultimately completes the task required of it, the situation has raised many questions within the community. This December 8, OpenAI finally spoke out, affirming that the situation is not intentional, the chatbot having not been updated since November 11. The firm is now trying to resolve the problem.





Source link -101