Last year, Ubisoft unveiled its ‘Commit’ artificial intelligence, an assistant intended to spot bugs before they make it into a game’s final code. Keen to show that its machine learning efforts have applications beyond the video game industry, the company has now teamed up with Mozilla to detect errors within the Firefox updates before they go live.
Like a video game, building a browser is no small feat. As Mozilla states, the application needs to play audio-visual material, handle network protocols and “secure communications using advanced cryptographic algorithms.” Millions of lines of written code make up the browser you’re reading article this from, and just one mistake can cost hundreds of hours to fix and a less-than-smooth experience for the user.
Although Ubisoft touted a 60-70 percent detection rate at the time of release, it simultaneously had a 30 percent false positive rate. Fortunately, Commit is based on Ubisoft’s CLEVER research program which utilises machine learning to gradually lessen this high false positive rate. It has not yet been revealed how effective CLEVER has been within its first year.
Considering Mozilla releases a Firefox update every “6 to 8 weeks,” this is sure to take a load off the engineering team when the newly dubbed Clever-Commit is fully integrated into the code writing process. Mozilla will begin using the tool during its code review phase, before eventually implementing it into other stages of code writing if proven successful.
“We expect to save hundreds of hours of bug riskiness analysis and detection,” reveals the company. “Ultimately, the integration of Clever-Commit into the full Firefox developer workflow could help catch up to 3 to 4 out of 5 bugs before they are introduced into the code.”
KitGuru Says: Considering browsers handle an awful lot of personal data, it’s good to see Mozilla doubling down on its bug fixing initiative. Ubisoft’s Clever-Commit is sure to have even more applications than we’ve seen, but it’s still early days for the year-old machine learning tool.