Few industries change as rapidly as gaming.
Studios are under constant pressure to develop games with better graphics, more intuitive mechanics, and larger maps. At the same time, those studios still have to deliver great storytelling, realistic dialogue, and soundscapes that are immersive to players.
To achieve this, major studios employ large teams of multitalented people — for context, Rockstar employed around 3000 people during the production of Red Dead Redemption 2.
Talented humans will always be involved in the development of video games. But, as technology develops further, studios will have a new tool in their inventory: Artificial Intelligence. AI is becoming more powerful, and deep learning programs can change the way games operate forever.
AI has been used in gaming for a while. It’s hard to pinpoint the exact moment when AI and gaming collided, but most folks caught on to the potential of AI when IBM’s Deep Blue program beat the chess world champion, Gary Kasparov, in 1997. Things have moved quickly since ‘97, and almost all video games use some limited form of AI to generate responses to players’ behaviors.
At its most basic, AI is used to present challenges to players in video games. This can mean anything from life-like opponents in FIFA and Madden or foes who react to player decisions in games like Demon Souls or The Witcher 3. AI in video games also produces some of our favorite features like “stealthing” in Assassin’s Creed — where players must remain undetected by AI to creep past or assassinate targets.
However, so far, video games only make use of “weak AI” — AI that has a predetermined set of behaviors based on a pre-programmed “decision tree”. This means that the AI in current video games is not “learning” and making decisions in the same way that humans do, but is following a pre-written script with a few variables, instead.
That is set to change, as developers are poised to start using “strong AI” in their games which utilize deep learning to generate responses that mimic human intelligence and draw from a far wider range of inputs and data.
Gamers have been promised the “next generation” of non-playable characters (NPCs) for a while now. And for good reason: few things break players from their immersion more often than glitchy or unrealistic NPC characters. Some games like Red Dead Redemption 2 and the Last of Us 2 have succeeded in creating mostly-realistic NPCs, while others, like CD Projekt’s CyberPunk 2077 woefully undelivered on their promise of realistic NPCs.
But even games with good NPC mechanics have not harnessed the true potential of AI — yet. Major players like Microsoft and EA are currently investing time, money, and effort into developing software that uses AI to generate new NPCs, as well as deep learning and reinforcement learning that ensures those NPCs are as “intelligent” as real players.
Listening to AI developers gives you the sense that we are on the brink of next-gen NPCs. Rockstar has already patented its AI-driven NPC system, and it’s not unreasonable to believe that games like GTA 6 and Elder Scrolls 6 will start to make use of the emerging tech to create worlds that feel real and NPCs that move beyond scripted decision trees and become more “human” than ever before.
Almost all gamers have finished a game only to wish it wasn’t over. We learn to love the “world” of games like Skyrim, the Witcher, and Horizon Zero Dawn. Currently, we have to wait for DLC content to be released that expands the games we love further, or simply restart the game and start the story all over again. But AI might change that.
We already have a taste of what AI-generated storylines might look like — Skyrim “randomly” generates quests for players after they’ve finished the main storylines, so players can theoretically go on questing in Tamriel forever. But, even this random generation of quests doesn’t come close to the potential of AI in storyline generation.
This is all speculative, but AI may be able to generate additional storylines based on your decisions in RPG games and the way you impacted the world during your playthrough. So, for example, if an AI learns that you tend to choose conflict over diplomacy in a future edition of the Witcher series, then perhaps you’ll find yourself running into the consequences of those decisions. In effect, this will generate a unique gameplay experience, as every player will have their own “world” that is created through their individual actions and deep learning programs.
However, AI-generated storylines do present an ethical quandary. Left to run amok, AI can generate storylines that are potentially traumatic for players and cross moral and ethical lines. This occurred recently when an AI-powered DND game called AI Dungeons produced content that was simply harmful and inappropriate.
The future of AI-generated storylines will need some oversight from human developers who can make judgment calls about the kinds of storylines and features that AI produces. Nonetheless, a future of AI-powered stories promises to give us gameplay experiences that last longer, and are more immersive.
The development opportunities offered by AI programs are exciting and should improve all aspects of player experiences. However, we mustn’t forget that AI is also a brilliant sales tool. That’s because deep learning programs use data sets that would be incomprehensible to even the most intelligent of humans — and this will be used to sell.
It may be cynical, but microtransactions are here to stay. After all, gaming companies make millions from microtransactions, and AI could significantly improve their offering. It’s not hard to see the potential, either: you get stuck on a boss or link your gaming profile to your social media pages, and suddenly downloadable content that “suits” you and your appearance is generated in the store.
Hopefully, AI-powered microtransactions won’t be as invasive as microtransactions in games like Ubisoft’s Assassin’s Creed: Valhalla, where having to toggle through paid content leaves players feeling exploited and can derail the gamer’s experience.
AI-powered video games will be fun — that’s a given. But, AI gaming also promises to change the world on a larger scale, primarily through education. Educators are already leveraging gaming in the classroom and the workplace. Currently, gamified learning experiences are not particularly complex, but AI-powered video games can take them further by tailoring the experience to each child on an individual basis to provide immersive AI-guided learning.
Immersive, gamified learning could take many forms. Entire games could be created that support learning through avatars and characters, and you can almost imagine the potential of open-world educational games which could engage students who seem disinterested in the content.
For example, if a student struggles with STEM content, then AI-powered education games can draw from previous examples of disinterested learners to spark an interest in STEM by using learning materials to present problems, real-world examples, or creative challenges in a way that engages that individual student.
For now, these kinds of immersive educational experiences are more dream than reality, but as deep learning programs become more sophisticated, it’s entirely possible to imagine that AI can help learners overcome the barriers to tricky content that has put children off of subjects like math and engineering for decades.
AI has massive potential in the gaming industry. We all want more immersive, realistic games that ditch decision trees in favor of human-like responses. Fortunately, we may not be that far from a future free from buggy NPCs who make dubious pre-scripted decisions. It’s hard to know exactly how this will look, but one thing is certain: the developers who can use strong AI in their games are sure to make massive profits on their investment.