Artificial intelligence has been one of the most revolutionary technologies in recent years. From task automation to content creation, AI has proven to be a breakthrough for people around the world.
However, Like any technology, it is not perfect and there is still a lot of work to do to improve it..
Recently, it has been reported that ChatGPT, OpenAI’s more than famous AI-based chatbot, has been providing users with formulas and guidance on how to perform tasks instead of completing them.
Although ChatGPT is known for its ease of use and always or almost always giving you the keys to what you need, Users have reported that the chatbot has become surprisingly lazy recently and OpenAI has assured that it is investigating this behavior to be able to resolve it, since it considers that it is not normal.
Bugs in ChatGPT that it seems that OpenAI wants to solve immediately
User comments on Reddit and Twitter suggest that ChatGPT responses have become somewhat limited. When they ask ChatGPT to perform a simple task that requires development time, they say it gives them a guide to the task instead of doing the work..
For example, a user posted a request they made to the chatbot on X. He asked her to extend him a line of code. Instead of doing it, since, as he explains, it should be about 50 lines of code, he gave him an initial guide on how to do it, along with an example, to tell him at the end that he continue pulling that thread to complete what he I was looking for.
“Sometimes I ask ChatGPT to parse a pdf or text file and it will do it quickly and smoothly. Other times, it will split my request into multiple messages. For example, it will only do half the parsing and then say “should I continue?” “.So I have to waste another message. Has anyone else noticed this? Other times, it will just flat out say that the task is too “computationally intensive” or something, even though it has done it before without problems,” explains another.
In the end, If the user has to do the work 100%, then the usefulness of artificial intelligence like these is quite limited.
