Home / Tech News / Featured Tech News / Ubisoft is using machine learning to spot bugs before they make it into the final game code

Ubisoft is using machine learning to spot bugs before they make it into the final game code

Ubisoft has had its issues with buggy game launches in the past and while things have improved over time, this has come at the cost of development time, as play tests get underway. In the future, this may not be an issue though, as the publisher has developed an ‘AI assistant' for its developers, which uses machine learning to catch bugs before they even make it into the final game code.

Ubisoft unveiled its ‘Commit Assistant' at a recent developer conference in Montreal, with the goal of saving developers time and reducing the number of potential bugs in a final game. To achieve this, Ubisoft's R&D division partnered up with Canadian universities and fed the assistant with ten years worth of game code from Ubisoft's library. Thanks to machine learning capabilities, this allowed the assistant to learn where mistakes have been made, what corrections were made and predict future mistakes developers might make.

As Ubisoft's video on the subject points out, the Commit Assistant is currently capable of spotting 6 out of 10 coding errors. Currently, the assistant raises a false alarm 30 percent of the time, but over time, this will be reduced as more code is fed through the algorithm.

Once a coding error has been spotted, the assistant can zero in on the most likely cause and suggest fixes for programmers to implement. In the long run, Ubisoft reckons this will save developers around 20 percent of their time. Speaking with Wired, Ubisoft's R&D head, Yves Jacquier spoke a bit about how this will impact the inner workings of Ubisoft's game studios:

“The most important part, in terms of change management, is just to make sure that you take people on board to show them that you're totally transparent with what you're doing with AI – what it can do, the way you get the data. The fact that when you show a programmer statistics that say ‘hey, apparently you're making a bug!', you want him or her to realise that it's a tool to help and go faster. The way we envisage AI for such systems is really an enabler. If you don't want to use that, fine, don't use it. It's just another tool.”

We don't know how this is impacting Ubisoft's current game projects, as the Commit Assistant is only just starting to be introduced. However, this is certainly an intriguing idea. It will be interesting to hear what developers think once the system is put into proper practise.

Discuss on our Facebook page, HERE.

KitGuru Says: Using machine learning techniques to catch bugs before they make it into the final game code sounds like a good idea, as long as the system is accurate. Crunch is a very real problem in the Triple A game development space, so this could potentially help alleviate some of that. 

Become a Patron!

Check Also

Wolverine Creative Director joins Xbox to lead Perfect Dark reboot

The Initiative has signed up former Wolverine creative director, Brian Horton, to lead work on the new Perfect Dark game.