Patent Application for Large Language Model Code Translation Error Detection

Strategic Insights -

Large language model code translation error detection technology has taken a major leap forward with a recent patent application filed by inventors Michael Gagliardi, Andrew C. M. Hicks, and Ryan Lo. The application, published on July 3, 2025, proposes innovative methods, apparatus, and products for identifying potential errors in code translation using large language models.

The inventors describe the challenges of migrating legacy code to more modern programming languages, which can lead to disastrous outcomes if deployed with errors. To address this issue, their patent application outlines a method for large language model code translation error detection, which includes receiving a code portion, converting it to a second programming language, calculating the accuracy of the conversion, and determining the difference between the accuracy and a historical accuracy. If the difference exceeds a predetermined value, the method indicates a potential error in the code portion.

The application also covers various aspects of the method, including determining the historical accuracy based on previous conversions, converting the code portion using a generative artificial intelligence model, and providing an indication of potential error to the model. Additionally, the patent application includes apparatus and computer program product claims that perform the method.

This breakthrough in large language model code translation error detection technology has significant implications for software developers and companies, as it enables them to improve system performance, increase maintainability and readability, and reduce the risk of errors in code migration.

Key Takeaways:

  • The inventors have filed a patent application for a method, apparatus, and products for large language model code translation error detection.
  • The method includes receiving a code portion, converting it to a second programming language, calculating the accuracy of the conversion, and determining the difference between the accuracy and a historical accuracy.
  • If the difference exceeds a predetermined value, the method indicates a potential error in the code portion.
  • The application covers various aspects of the method, including determining the historical accuracy based on previous conversions, converting the code portion using a generative artificial intelligence model, and providing an indication of potential error to the model.
  • The patent application includes apparatus and computer program product claims that perform the method.
  • The technology has significant implications for software developers and companies, enabling them to improve system performance, increase maintainability and readability, and reduce the risk of errors in code migration.

Statistics:

  • The patent application was filed on December 28, 2023.
  • The application was published on July 3, 2025.
  • The inventors, Michael Gagliardi, Andrew C. M. Hicks, and Ryan Lo, are experts in large language models and software development.
  • The method and technology described in the patent application have the potential to revolutionize the way software is developed and maintained.

Sources:

  • Gagliardi, Michael; Hicks, Andrew C. M.; Lo, Ryan. Large Language Model Code Translation Error Detection. U.S. Patent Application Number 20250217126, filed December 28, 2023 and posted July 3, 2025.
  • NewsRx LLC. "Patent Application for Large Language Model Code Translation Error Detection Published." NewsRx, 2025, https://www.newsrx.com/news-article/2025072312862296.html.