4 Ways AI is Changing Gaming for the Better

By 2026, the worldwide video gaming/eSports industry will be worth $295.63 billion, assuming it maintains its current growth rate, according to technology forecasts. Given that games are perfect for a world that’s still social-distancing, it’s no wonder that game developers are interested in leveling up the gaming experience with artificial intelligence and machine learning (AI/ML) models. AI/ML (and the Samsung memory technology enabling AI/ML) are already changing game play through image upscaling and offering data-driven, personalized in-game purchases. In the future, we can expect AI to deliver significantly smarter chat bots and even truly intelligent games. 1. Happening Now: Upscaling Images As any gamer will tell you, increased graphic fidelity can drastically improve overall gaming experience, making play feel more realistic and immersive. That’s why NVIDIA created an artificial neural network — to teach select gaming systems to enhance images during game play through a process known as upscaling. Here’s how image upscaling works in gaming systems (like NVIDIA’s GeForce RTX 20 and 30 Series): 1. First, an artificial neural network was developed and trained on NVIDIA’s DGX servers. These are powerful, data center-based servers that utilize Samsung High Bandwidth Memory. 2. The neural network is shown millions of images and taught to recognize “ideal” (hi-res) images that correspond to low-res ones. This training enables intelligent upscaling — the efficient recreation of life-like images using significantly lower resolution ones. 3. DLSS is installed on the GeForce RTX systems, which use dedicated AI processors (and also Samsung’s GDDR6 SDRAM). 4. During game play, the intelligent upscaling amps the resolution of native game visuals. Combined with a bumped up frame rate, game play goes from meh to mesmerizing. While the GeForce RTX systems are leveling up the standalone gameplay experience, AI/ML innovation is also reshaping the online gaming space.

2. Happening now: Personalized in-game purchases While many online games are free to play, studios up the experience through in-game/in-app purchases, and player-specific, targeted advertising. By leveraging AI-powered data mining, studios are able to analyze game-play data to predict what a player might purchase and when they might be most apt to purchase it. Data mining can not only help studios increase industry revenues but also to better understand players’ experiences and develop more in-depth player profiles. Seemingly simple things, like easily getting your favorite skins, or the booster you need to beat a tough level, can be key to both studio success and player loyalty.

The amount of player feedback data that needs to be stored and analyzed to create intelligent recommendation engines and personalized gaming purchases is huge, but with the help of AI, this data can be leveraged to the benefit of game developers while helping to enhance the player experience. 3. Soon: Better Conversations with Natural Language Processing Natural language processing (NLP) is another compute-heavy technology in which AI/ML is being used to improve user experience. NLP research goes beyond the world of video games, of course, but its application within games would go a long way toward deepening the immersive aspect of play. As anyone who has ever tried to learn a second language knows, natural speech can be ambiguous, contradictory, and difficult to parse. Comprehension can hinge on nuance, cultural knowledge, tone, and facial expressions/physical gestures as much as on spoken or written words. And there are technically an infinite number of possible utterances in any given language, not including hybrid languages like creoles. Naturally, the compute required for NLP is enormous and depends on artificial neural networks to process and manage data based on hundreds of billions of parameters—that’s where Samsung HBM (High Bandwidth Memory) comes in.

Samsung HBM includes super-fast, stacked memory die to move mountains of data to and from the processor. Given the high-performance computing required to teach computers to understand human language, Samsung’s HBM is the go-to memory for data centers training artificial neural networks for NLP. 4. Stay Tuned: Bespoke gaming If you’re of a certain age, chances are that when you think of AI and gaming, you think of IBM’s Deep Blue besting chess legend Garry Kasparov in 1997. And indeed, many AI gaming algorithms focus on the logic behind games like chess and Go. But when it comes to role-playing games (RPGs), where a distinct plot is developed, AI and ML have not typically been used to propel narrative; not only have the memory and processing requirements been too expensive, but also, a certain amount of predictability is purposely built into games. After all, a game that is too hard to predict wouldn’t appeal to most human players, who want to at least have a fighting chance at winning. As such, game story arcs have traditionally been determined by decision trees, some of which are very sophisticated; but after repeated game play, non-playable characters (NPCs—that is, characters created and played by software) can end up being too predictable. So, instead of using AI to build an ever-morphing, unwinnable game or super-smart NPCs, studios are beginning to explore the kind of deep learning needed to create intelligent, adaptive games. These personalization algorithms would not only allow software to tailor games to individual players, accommodating player skill level, playing style, style, weaponry, etc., but even to design—or potentially help players design—entirely new games from scratch on the fly. NPCs could adapt to suit a players’ personal preferences for language use and attitude. These advanced capabilities may be a few years into the future, but Samsung’s HBM and graphics DRAM solutions will be involved in offering the performance, capacity and efficiency required to support next-generation AI and ML capabilities, in the world of gaming and beyond.