JPMorgan Patent Would Double-Check AI-Generated Code

Group 1 - JPMorgan is seeking to patent a method for evaluating the quality of software code generated by a large language model (LLM) [1] - The process involves a user providing instructions and the LLM selecting an API to write the code, which is then executed to verify its correctness [2][3] - The system assesses the code's quality based on accuracy, reliability, robustness, and consistency [3] Group 2 - JPMorgan has previously sought patents to ensure accuracy, exploring machine learning and blockchain for fact-checking [4] - Other tech companies, such as Microsoft and IBM, are also addressing AI-related issues, including code generation accuracy and AI bias [4]