Since its launch in November 2022, the AI tool called ChatGPT has become deeply involved in the lives of many people worldwide.
It is practiced in the net and tech world due to its ability to answer various questions in realtime, from essays to writing mails, to arithmetic. It is even feared that in the near future people will lose jobs by replacing people in various fields. But recently the accuracy of ChatGPT chatbots has again been questioned. According to media reports, ChatGPT gave 52 percent incorrect answers to software engineering questions.
Is ChatGPT really worthy of such popularity?
According to a recent report, a
study by Purdue University in the US has questioned the accuracy of chatbots.
In this case, researchers at that institution asked ChatGPT 517 questions about
Stack Overflow (SO). And this is when the AI tool gives 52 percent answers or
analysis errors. Not only that, 77% of the answers given by ChatGPT were found
to have wrong verbs. However, it is said to have given the wrong answer due to
not understanding the underlying concept of the question.
Naturally, researchers have
questioned the limited reasoning power of these AI tools regarding their
answers. According to them, ChatGPT can be helpful for understanding solutions
to problems such as quick engineering and human-in-the-loop fine-tuning, but
falls short when it comes to injecting logic into LLM. So they feel it is very
important to correct it.
The use of ChatGPT is gradually decreasing An earlier report found that the number of ChatGPT users has been declining rapidly in recent times. And this trend is still continuing. Last June, 1.7 billion people were using the tool, but in July this number dropped to 1.5 billion. It has even been said that OpenAI is not getting profitable results. The company is reportedly spending around $700,000 per day to operate ChatGPT. However, it has also launched its mobile app a few days ago. All in all, if this continues, ChatGPT may soon go bankrupt – it's safe to say!