Forty years since PAC-MAN's glory days in the arcades of Japan (and later throughout the world), the most classic video game on Earth is reborn with the help of AI, according to a post on NVIDIA's website.
RELATED: GENERATIVE ADVERSARIAL NETWORKS: THE TECH BEHIND DEEPFAKE AND FACEAPP
PAC-MAN is back thanks to AI
The powerful new AI model — courtesy of NVIDIA — trained on 50,000 iterations of the game, and can generate a fully functional emulation of PAC-MAN without an underlying game engine.
This is significant because it means AI can recreate a convincingly complete game without ever receiving programming about the game's fundamental rules.
GameGan is the first neural network model to successfully mimic a computer game engine with generative adversarial networks (GANs). These involve two neural networks in competition — a generator and a discriminator — which work to create new content convincing enough to fool us, or more importantly the discriminator, into recognizing it as the original.
"This is the first research to emulate a game engine using GAN-based neural networks," said Seung-Wook Kim, a researcher at NVIDIA and lead author on the new GAN-based 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."
GAN: Generative Adversarial Network
While an artificial agent plays a GAN-generated game, GameGan reacts to the agent's developing actions, generating new frames of the game's environment in real-time. GameGan even generates novel game layouts never before witnessed by the program if it's trained on screenplays from games with various versions or multiple levels.
This new capability might be useful to game developers looking to auto-generate layouts for new game levels, in addition to AI researchers interested in finding ways to lower the difficulty of developing simulator systems for training autonomous machines.
"We were blown away when we saw the results, in disbelief that AI could recreate the iconic PAC-MAN experience without a game engine," said Koichiro Tsutsumi of BANDAI NAMCO Research Inc., the company behind the game's publisher BANDAI NAMCO Entertainment Inc. that provided PAC-MAN data for GameGAN's training, according to NVIDIA's post. "This research presents exciting possibilities to help game developers accelerate the creative process of developing new level layouts, characters and even games."
The new AI-based PAC-MAN game will be available later this year on AI Playground, where anyone can try out the new GameGAN PAC-MAN system for themselves.
As artificial intelligence moves forward in engineering, social media, entertainment, culture, and even art — there's little to say about the future of media, without keeping AI and GAN on our collective dashboards.