A GPT-4-based model called 'CriticGPT' will be developed to detect mistakes made by ChatGPT
OpenAI has announced that it has developed an AI model called ' CriticGPT ' that detects errors in ChatGPT. Like ChatGPT, CriticGPT is based on GPT-4.
Finding GPT-4's mistakes with GPT-4 | OpenAI
Chat AI such as ChatGPT allows you to generate code and create long sentences with minimal effort. However, the code and sentences generated by chat AI often contain errors, and there have been reports of users using code generated by ChatGPT as is, only to discover bugs that caused actual harm.
A story of failure that resulted in a loss of over 1.5 million yen due to an overlooked bug in the code generated by ChatGPT - GIGAZINE
OpenAI has developed a new AI model called 'CriticGPT' that detects and corrects errors in ChatGPT. CriticGPT is a model developed based on GPT-4, and its code error detection and correction capabilities have been enhanced by learning 'code that manually includes errors' and 'sentences that correct code errors.'
Below is an example of how to use CriticGPT. In response to the code generated by ChatGPT, CriticGPT points out that 'using the startswith method for this purpose is not appropriate' and offers an alternative.
The graph below compares the completeness of code critiques by humans (green), CriticGPT (orange), and humans and CriticGPT (pink). CriticGPT critiques are more complete than human critiques.
Below is a graph comparing the 'percentage of false information (
OpenAI states that 'better tools are needed to adjust increasingly complex AI systems,' and intends to continue developing tools to adjust AI output such as CriticGPT.
Related Posts:
