Nvidia researchers have successfully used a generative adversarial network AI called GameGAN to produce a fully functional version of the 40-year-old game PAC-MAN.
GameGAN was trained through watching 50,000 PAC-MAN episodes, after which the AI become capable of recreating the classic game – without an underlying game engine. GameGAN is built using a Generative Adversarial Network, or GAN for short. According to AI pioneer Yann LeCun, who oversees AI research at Facebook, GANs are “the most interesting idea in the last 10 years in machine learning.”
Video credit: Nvidia
“This is the first research to emulate a game engine using GAN-based neural networks,” said Seung-Wook Kim, an NVIDIA researcher and lead author on the project. “We wanted to see whether the AI could learn the rules of an environment just by looking at the screenplay of an agent moving through the game. And it did.”
But the research isn’t just aimed at reproducing games, as Nvidia currently employs 200 scientists focused on AI with the hope that one day it will help out with other areas such as “computer vision, self-driving cars, robotics and graphics.”
Nvidia will make the AI created game available later this year on AI Playground, where anyone can experience the game firsthand.
More details from Nvidia relating to GameGAN can be read HERE.
Discuss on our Facebook page HERE.
KitGuru says: What are your thoughts on a computer being able to recreate a game by only observing its gameplay?