Video games can be much more than a form of entertainment. And for the auto manufacturers of the world who need data on how driverless cars should behave, one game in particular is proving to be quite useful. That game? Grand Theft Auto 5.
In a recent interview with Bloomberg, a Princeton University professor of operations research and financial engineering who advises the Princeton Autonomous Vehicle Engineering team called GTA 5 “the richest virtual environment that we could extract data from.”
You might be thinking, “Wouldn’t it be better to just send cars into the real world?” And to be fair, researchers are doing that – but GTA 5 and other software can produce data much faster and without the need for all the hiccups that come with a real car, like gas and maintenance, so it makes for a good supplemental tool. And it’s not like they’re gathering the data from the reckless GTA Online players of the world – these are simulations, not for-funsies play sessions.
Still, it reminds me of the “Corrupted Blood incident” that happened in World of Warcraft back in 2005. For those unfamiliar, an in-game disease called Corrupted Blood could be cast on players by the final boss of the Zul’Gurub raid.
The sickness drained players of hit points, and was highly contagious. While it was designed only to last a few seconds and be limited to the raid in which it originated, players found ways to spread it to the outside world. This caused a world (of Warcraft)-wide pandemic that was studied by the CDC as a model for how contagion can spread.
So yeah, I can see how GTA 5, with its plethora of car models, unpredictable AI, and multiple types of weather, could be useful in generating data. It’s not like the algorithms created from said data are going to turn your car into a local crime boss or make it catch fire and explode within seconds of flipping over. At least probably not.