Engineer Davide Bacchet with the self-driving startup NIO told Bloomberg that it's just "not practical" to rely solely on the public tests to teach their software everything it needs to know to rise to an acceptable standard, adding that the ability to replay the same exact scenario over and over again is key for learning the correct course of action. Simulators are becoming a key part of autonomous vehicle development as red tape and legal challenges have limited the spread of real-world testing to just a few states. In fact, one researcher with Princeton University called GTAV "the richest virtual environment that we could extract data from." Accumulated over the course of millions of in-game miles driven, the resulting bounty of information would otherwise take years to collect on real-life public roads.
But according to Bloomberg, rather than emulating a caffeine-crazed teen crashing through traffic, these firms are taking advantage of the game's incredibly detailed environment to run traffic, weather, and accident-avoidance simulations. Waymo and Toyota Research won't be turning their autonomous cars loose on the streets of Los Santos to maim and murder innocent civilians any time soon. So it's understandable if you're maybe just a little bit concerned at the news that some of the top self-driving car developers are turning to GTAV to power their simulators and teach their software the virtual rules of the road. Whether you're level 250 in the latest iteration of GTA Online or you haven't picked up a controller since the PlayStation 2 days, chances are your last visit to Rockstar's chaotic world involved a whole lot of automotive hi-jinks that would likely earn you a quick trip to the slammer in real life. Think about the last time you played a Grand Theft Auto game.