How Artificial Intelligence is Changing Online Gaming and Game Development

Artificial intelligence holds great promise to improve some of the most important aspects of human society: healthcare, the economy, transportation, engineering… it will be a game-changer in every yardstick by which progress is measured for humanity.

But it may also just be a lot of fun.

Already, the most pervasive and popular use for generative artificial intelligence is simply having a conversation with a machine. Chatbots inspire and delight. They create whimsical pictures and silly poems. They are entertainment.

Entertainment is also a business, though. When it comes to computer gaming, it’s an industry estimated to have generated revenues of nearly $190 billion in 2023. Some 3 billion people play regularly, or just over 40 percent of the world’s population.

That’s a lot of people having a lot of fun.

AI has the potential to entirely revolutionize their experiences on every level. New worlds of enjoyment are about to be ushered in by the machines.

Artificial Intelligence Is Old School in Computer Game Development

Video game AI has been a long-term bright spot in the field of artificial intelligence.

While the rest of the AI industry hibernated through a long winter bare of research funds or interest, the gaming industry was going full bore on techniques to make computerized opponents or characters seem smarter, more real, and more capable.

AI is used in games to provide competing players, to create life-like responses to players actions, and to create believable situations and environments on the far side of the computer screen.

Even very basic games make use of algorithms to drive their loop. A chess program has to be able to simulate a human player on the other side. Massively Multiplayer Online Role-Playing Games (MMORPGS) need AI to provide movements and actions for monsters and non-player characters.

Gaming AI hasn’t been closely associated with academic AI because it focuses on the player’s experience rather than the path to get there. Game programmers don’t care if the AI is actually smart. They care if the player believes that it is. A combination of clever coding and an understanding of human psychology could be as good as a legit machine intelligence breakthrough.

Modern developments in AI owe a lot to games.

But games have turned into a proving ground for academic AI concepts. Teaching AI how to play everything from chess to Go has resulted in a lot of breakthroughs in the modern AI industry.

With new capabilities coming from generative AI in the academic world, game AI engineers and players alike might start to reap the benefits of artificial intelligence.

How Breakthroughs in Artificial Intelligence Will Completely Change Game Design and Construction

One interesting feature of AI in games is that limitations have spurred innovation. Development has often been constrained by the technical restrictions of coding for home computers or game consoles. With a lot less horsepower to throw at problems, game programmers developed every trick in the book to get realistic performance out of their code.

Game programmers relied on technical tricks to make their AI seem more lifelike. Basic branching algorithms made game AI seem as if it were considering different options for a play-through; random pathing could make opponents seem like they were hunting.

With big companies like Apple already working out ways to cram algorithms down into versions that can be powered on phones, more solutions are on the horizon.

Innovations like procedural generation introduced randomness into games. Behavior trees, finite state machines, and other elementary AI techniques could deliver apparently complex behaviors with relatively little computational cost.

Generative AI is a perfect fit for packing character interactions with realism. Current games rely on scripted conversations, written and performed by human voice actors. But that limits interactions and can make for dialogue that’s stiff and formal.

Artificial intelligence may also find its way into board game design, generating art assets and checking game balance for designers even if the game won’t be played on computers.

Natural language processing can make in-game conversations flow any direction the player wants to take them. And with AI voice generation and recognition coming along quickly, those conversations may well be entirely in audio form.

Generative AI can also quickly generate game assets, such as artwork, music, or description text to fit any design. In fact, AI is already being used to smooth animations to reduce workload on human animators and the need for intensive motion capture.

AI Engineers in Gaming Need to Help Machines Find the Fun in Games

While AI can come up with conversations, play Go, or wipe the floor with human players in Unreal Tournament, it doesn’t yet have any idea how to make things fun.

One thing that makes games against human players fun is that those players learn; try a trick move twice in a row and you’ll get skunked. But most programs aren’t that smart—the same moves work against them no matter how often you use them.

Having fun is one area where machine learning can let the machine, well, learn from the player.

In fact, data science and machine learning have already been heavily used in game development to analyze player data in beta testing. Data scientists develop heat maps and analyze play-through instrumentation to find parts of the game that are too hard, or too easy, or that players prefer to repeat.

Future games may incorporate ML algorithms that skip a step. Rather than being trained by developers after their analysis, the algorithm may simply learn directly from what the player does in the course of gameplay. That can occur on the story level as well as the tactical level… a game might figure out what sort of challenges you enjoy and seed the progression with more of those. A different player, in the same game, might get a completely different set of events.

AI in Gaming May Bring Human Players Closer Together Than Ever

Multiplayer is another area where AI seems destined to shine… and not just as more realistic bots to fill empty arenas. Instead, game companies are likely to use AI to manage in-game support, scan for cheating players, or smoother interactions between current players. In fact, Roblox, a popular platform hosting multiple multiplayer games with more than 70 million users worldwide, recently rolled out a live AI-powered translation service.

Using a specially trained LLM versed in 16 languages, the system instantaneously translates from a user typing in their native language to the languages of other players across 180 countries. With more than 2.4 billion chat messages exchanged daily, the service dramatically opens up the ability to meet new people around the world and tell them they’ve been owned.

Similar technology may be able to instantly localize games originating in any language, expanding markets for game publishers quickly and inexpensively.

What Kind of Jobs Will Emerge in the Gaming Industry for AI Professionals?

Many game developers have a leg up on professionals from other industries making the transition to AI. So many games incorporate various kinds of AI technology already, it’s not a big leap to move from procedural generation to full-on generative systems.

On the other hand, gaming is such a specialized type of software development that it can be challenging for pure AI engineers to make the jump to game dev jobs.

Most of the positions in the industry will be scattered among the current types of game development studios in positions such as: