AI Predicts Coding Mistakes Before Developers Make Them
Ubisoft's new AI tool could make it easier and cheaper to code video games.
Ubisoft, the French video game company, debuted a new artificial intelligence (AI) that proactively flags coding bugs at its recent developer conference in Montreal. The tool, named Commit Assistant, is intended to catch mistakes before developers even commit them in a game’s code.
The company fed approximately 10 years’ worth of code into Commit Assistant from across Ubisoft’s library of software, according to a report from Wired U.K. The AI analyzed where previous mistakes were made in the code and what corrections were applied to rectify those mistakes. This type of machine learning enabled Commit Assistant to predict when a programmer could be at risk of introducing a similar bug.
Audiences expect that video games released by major publishers like Ubisoft would be free from bugs. However, eliminating every single bug from these products can be laborious and time-consuming. GamesIndustry.biz reported that the company’s research and development head, Yves Jacquier, said the AI could save programmers 20 percent of their time. Ubisoft also claimed that Commit Assistant can catch six out of 10 bugs accurately.
If developers don’t catch coding mistakes at their inception, it might fall to paid video game testers to point them out post-production. Once they’ve highlighted an issue, it then falls to the development team to figure out what chunk of errant code is responsible — a massive undertaking. According to Wired U.K., Ubisoft said catching an error before the game hits the shelves can absorb as much as 70 per cent of costs of eliminating bugs.
While Ubisoft’s version of the tool is geared toward video games, the same concept could potentially scope out bugs in all kinds of software development projects. Commit Assistant was created in partnership with the University of Concordia, and the school will soon publish academic papers on the underlying methodology, which could help other software developers utilize the same type of machine learning to streamline their coding process.
However, this AI’s success was in part due to the massive amount of data Commit Assistant was able to pore over before offering bug predictions. A multibillion-dollar international company like Ubisoft has enough raw information to train this type of predictive AI effectively, but smaller developers will likely not be able to match the same results.
As a Futurism reader, we invite you join the Singularity Global Community, our parent company’s forum to discuss futuristic science & technology with like-minded people from all over the world. It’s free to join, sign up now!