The JPMorgan patent will double-check AI-generated code

AI can always write for you, but it is not always correct.

That’s why JPMorgan wants to double-check your LLM homework. The company is seeking a patent for the invention “A method and system for obtaining an evaluation of the quality of software code” Created by LLM.

Read also: Economists are optimistic about US growth despite rising inflation and a shrinking labor market and Broadcom’s mysterious $10 billion client highlights swelling AI investments

First, the user provides instructions for a task, including the output they are looking for. The system sends that information, along with a list of available APIs, to the LLM. The LLM is required to choose just one API and write code using it.

LLM then responds with the API it chose, as well as the code it wrote based on the instructions. The system then runs this code to see if the task ran correctly, with the expected result.

At the end of the process, the system checks the quality of the code: its accuracy, reliability, robustness and consistency.

JPMorgan has sought patents to ensure accuracy before. Last year, the largest US bank looked into the matter How machine learning and blockchain can verify facts. Accuracy has always been an issue that technology companies look forward to: Microsoft is also concerned about it Problems of creating artificial intelligence code He wants to target AI bias, and IBM is trying to use AI to fix AI problems.

This post first appeared on Daily uptrend. To receive razor-sharp analysis and perspective on all things finance, economics and markets, sign up for our free website Daily uptrend Newsletter.