Dec 5, 04:30 PM

ChatGPT's Code Checking Flawed, New Features Like GPT-4.5 and o1 Pro Mode in Development

A new study has revealed that ChatGPT, a popular AI tool developed by OpenAI, significantly overestimates its own ability to check the code it generates. Despite its capabilities in generating and debugging code, the large language model often misclassifies errors or misses them entirely. Meanwhile, OpenAI appears to be working on new features, including a 'limited preview of GPT-4.5' for the ChatGPT Team plan, which has not yet been made visible to users. Additionally, references to 'o1 pro mode' have been found in the code, suggesting an upcoming enhancement to the AI's capabilities, although details on its functionality and availability remain unclear.

#ChatGPT #OpenAI #ChatGPT Team

Written with ChatGPT .