52% incorrect answers by ChatGPT's on software engineering
IANS 14 August 2023
OpenAI's ChatGPT answered about 52 per cent software engineering questions incorrectly, according to a study, raising questions about the popular language models accuracy.
 
Despite ChatGPT's popularity, there hasn't been a thorough investigation into the quality and usability of its responses to software engineering queries, said researchers from the Purdue University in the US. 
 
To address this gap, the team undertook a comprehensive analysis of ChatGPT's replies to 517 questions from Stack Overflow (SO). 
 
"Our examination revealed that 52 per cent of ChatGPT's answers contain inaccuracies and 77 per cent are verbose," the researchers wrote in the paper, not peer-reviewed and published on a pre-print site. 
 
Importantly, the team found that 54 per cent of the time the errors were made due to ChatGPT not understanding the concept of the questions. 
 
Even when it could understand the question, it failed to show an understanding of how to solve the problem, contributing to a high number of conceptual errors, they said. 
 
Further, the researchers observed ChatGPT's limitation to reasoning. 
 
"In many cases, we saw ChatGPT give a solution, code, or formula without foresight or thinking about the outcome," they said.
 
"Prompt engineering and human-in-the-loop fine-tuning can be helpful in probing ChatGPT to understand a problem to some extent, but they are still insufficient when it comes to injecting reasoning into LLM. Hence it is essential to understand the factors of conceptual errors as well as fix the errors originating from the limitation of reasoning," they added.
 
Moreover, ChatGPT also suffers from other quality issues such as verbosity, inconsistency, etc. Results of the in-depth manual analysis pointed to a large number of conceptual and logical errors in ChatGPT answers. The linguistic analysis results showed that ChatGPT answers are very formal, and rarely portray negative sentiments.  
 
Nevertheless, users still preferred ChatGPT's responses 39.34 per cent of the time due to its comprehensiveness and articulate language style. 
 
"These findings underscore the need for meticulous error correction in ChatGPT while also raising awareness among users about the potential risks associated with seemingly accurate answers," the researchers said.
 
Disclaimer: Information, facts or opinions expressed in this news article are presented as sourced from IANS and do not reflect views of Moneylife and hence Moneylife is not responsible or liable for the same. As a source and news provider, IANS is responsible for accuracy, completeness, suitability and validity of any information in this article.
Comments
Send Anywhere: Transfer Large Files
Yazdi Tantra, 11 August 2023
This is a simple app to transfer files wirelessly. It is available on all major platforms. You can transfer files in real time using a six-digit key. You can also share files through a link, with multiple people or even send to your...
COOKmate: Your Personalised Cookbook
Yazdi Tantra, 04 August 2023
COOKmate is one of the best personalised cookbook apps on mobile. It lets you create your own digital cookbook by building your own database, gathering recipes on the web and importing them. You can search for recipes online and...
6000 Thoughts – AI Life Coach
Yazdi Tantra, 28 July 2023
This is a life coach app which helps you to organise your multiple thoughts into coherent data which you can use to take decisions. Sometimes, when you need a friend, philosopher and guide, simply pick up the app and speak out your...
Free Helpline
Legal Credit
Feedback