Collaboration will advance how modern enterprise software is secured and maintained
MONTREAL and MIAMI — Sept. 18, 2023 / BUSINESSWIRE/ — Mila - Quebec AI Institute, the world’s largest academic deep learning (DL) research center, has established a partnership with Moderne, the pioneer in next-generation automated code remediation and analysis. The collaboration brings together Mila’s expertise in artificial intelligence (AI) and DL and Moderne’s knowledge of rules-based refactoring systems to automate the work of securing, maintaining, and modernizing codebases at scale. Developers, who can spend at least 40% of their day modifying and updating existing applications, will benefit from this AI partnership.
“Our partnership with Mila will help Moderne to deliver the most effective way for humans and AI to maintain and secure source code together,” said Jonathan Schneider, CEO and co-founder of Moderne.
“Mila is excited to work with Moderne to explore AI for the tasks of code maintenance and security—some of the most manual, time-consuming, and tedious work developers face today,” said Stéphane Létourneau, Executive Vice-President of Mila. “We believe that companies in many sectors could benefit from the collaboration between our two organizations, which will aim to enable them to better visualize, analyze, and fix their code at scale.”
Moderne has hired Justine Gehring as a research engineer to advance its AI efforts and oversee its Mila collaboration. The company is particularly interested in using AI models to deliver more insights into today’s large, unwieldy codebases, such as ways to visualize similarities across an enterprise codebase and clustering code issues to enable more efficient actions. Moderne also will apply generative AI to accelerate the development of rules-based programs that are used to transform source code.
“Today, many products are using generative AI to support code authorship at the individual developer level,” Gehring said. “These models are useful but can include hallucinations, bugs, and insecurities. Using these models requires significant developer quality control and makes it difficult to scale. Plus, what about managing all the code being generated and modified now at this accelerated pace? Just imagine developers each making similar code fixes in thousands of places with AI support that must be individually scrutinized. This would take notable time per developer and could still result in inconsistent and error-ridden code.”
Code remediations are often multi-point operations that require real-time accuracy, consistency, and coordination across multiple repositories.
Schneider further explained: “Unfortunately, generative AI doesn’t scale to the level most organizations need for code security, maintenance, and migration. Moderne’s platform was built to specifically solve the problem of automated code remediation at scale. Now, by incorporating AI into our product, we can deliver the best of both worlds to our customers. Moderne is excited to partner with Mila to stay at the forefront of the rapidly moving AI tech space.”
Moderne initially will be co-leading Mila’s ML4Code (machine language for code) reading group, as well as continuously engaging in research with collaborators to advance the code security and maintenance use case. The ML4Code reading group explores and discusses the newest papers related to coding tasks, such as code generation, code maintenance, and code translation. These tasks are generally tackled using large language models (LLMs), such as StarCoder and Code LLama, and graph representations of code.
