Altran Technologies SA, a provider of digital design services owned by the French consulting firm Capgemini SE, today released a new tool for developers that uses artificial intelligence to predict the likelihood of bugs being found in software source code.

The tools is called Code Detect AI, and it works by applying machine learning to historical data in order to identify areas of code that are most likely to be buggy. From there, it suggests several tests developers can perform to diagnose and fix those flaws, resulting in much faster development times overall.

Altran, which developed the new tool in partnership with Microsoft Corp., said bugs are an inescapable fact of life when it comes to software development, so it’s critical that organizations can spot them as soon as possible. The later a bug is found, the more costly and time-consuming it can be to fix it. With Code Detect AI, companies can discover defects earlier and minimize the time it takes to fix them.

“It’s well-known that software developers are under constant pressure to release code fast without compromising on quality,” explained Walid Negm, group chief innovation officer at Altran. “The reality, however, is that the software release cycle needs more than automation of assembly and delivery activities. It needs algorithms that can help make strategic judgments ‒ especially as code gets more complex. Code Defect AI does exactly that.”

It’s trained using several machine learning techniques, including random decision forests, support vector machines, multilayer perceptron and logistic regression, to create a more reliable decision model, Negm said. The tool then scans each piece of source code that’s fed into it and provides a “confidence score” that indicates how likely that code is to contain bugs. In addition, developers can also use Code Detect AI to determine which features in their software are most likely to be responsible for causing bugs.

“When Code Defect AI discovers new developer commits, it predicts if any files in the commit are at risk for defects,” Microsoft explained. “The rationale behind the prediction is presented using Local Interpretable Model-Agnostic Explanations, so that developers can trust and learn from the prediction.”

Code Detect AI can be used to scan any code hosted on GitHub, the Microsoft-owned code hosting repository. It can also be used with other source code management tools, Altran said.

Photo: Elchinator/Pixabay