openai-unveils-criticgpt-to-improve-gpt-4

OpenAI has recently introduced a new AI model called CriticGPT, which is designed to detect and fix errors in code produced by GPT-4.

 

As described in a post on Thursday, the AI firm explained that CriticGPT utilises the reinforcement learning from human feedback (RLHF) framework, a method that blends machine output with human input to enhance AI systems.

 

CriticGPT is still in development and is not yet accessible to users or testers. Its objective is to improve the quality of AI-generated code.

 

OpenAI disclosed that during testing, individuals who used CriticGPT to review code produced by ChatGPT outperformed those who did not use the model 60% of the time. The research findings have been detailed in a recently published paper.

 

The RLHF framework involves human evaluators, known as AI trainers, providing feedback on the AI's performance to help adjust and enhance the model's behaviour.

 

Trainers deliberately introduced mistakes into code samples that already contained natural errors, and then provided example feedback on these mistakes. The model's ability to detect both naturally occurring and intentionally added errors was used to evaluate its performance.

 

As per OpenAI, CriticGPT showed a 63 per cent enhancement in identifying code errors compared to ChatGPT. Nevertheless, the model does have its limitations. Its training has mainly focused on short code snippets and it has not yet been tested on longer, more intricate coding tasks.

 

Furthermore, the model still struggles with producing inaccurate factual responses, an issue known as hallucination. Additionally, it has not been evaluated in situations where multiple errors are scattered throughout the code.