Generative artificial intelligence (AI) is perhaps the most relevant area of technology development today. Experts are actively looking for models of AI use in almost all industries, and video game development is no exception. However, the use of AI in game development conceals some potentially unpleasant consequences that could affect the entire video game industry.
Difference in terminology
Previously, when we talked about AI in video games, we usually talked about algorithms for the behavior of non-player characters. It was the AI rules that regulated how enemies, player’s partners, and other NPCs would behave. AI is still working in this field in games today, and the development of NPC behavioral rules has not changed much over the past few years and depends mainly on the imagination and talent of developers.
That’s why the battles in the good old F.E.A.R.: First Encounter Assault Recon or Half-Life 2 still feel very lively and vigorous even today. Because the developers have prescribed very good rules of behavior for the enemies to make the gameplay more interesting.
When talking about AI in games today, most people mean the use of generative AI, i.e. technologies that can generate some content (text, sounds, images, videos, etc.) based on a user request. There are already a lot of such services. ChatGPT, Midjourney, and Dall-E2 are just a few of the most famous examples.
The use of AI is gradually entering more and more strange territories. For example, the Oral-B brand has recently released a special toothbrush called Oral-B iO, which analyzes how you brush your teeth using artificial intelligence. It is able to provide “efficiency” indicators and indicate when you are putting too much effort into brushing (or vice versa – not enough effort).
Successful examples of AI use
Applying generative AI to game development seems like a rather obvious step. After all, studios work with a huge amount of diverse media content. AI can greatly simplify game development. Several notable releases have already proved this.
Take, for example, the humorous shooter High on Life, which was released in December 2022. Justin Roiland, whom you may know as the creator of the animated series Rick and Morty, was responsible for the creative part of the development. High on Life has adopted Roiland’s strange and provocative, but in its own way attractive humor.
It was the game’s creative director who admitted that the Squanch Games team used generative AI during development. Most of the art in the game was made by humans, but for some final touches, prototypes, or other small things, such as posters on the walls, the developers used artificial intelligence. Royland said that this allowed for a more “strange” atmosphere.
“AI images make the world feel like a strange alternate universe of our world. And we used it to come up with weird, funny ideas. I don’t know what the future holds, but AI is going to be a tool that has the potential to make content creation incredibly accessible,” says Justin Roiland.
In High on Life, AI was also used to create dialogues and NPC voices. The game’s chief designer, Eric Mair, stated that none of the fully AI-generated lines were used in the game. The technologies were used solely for testing and prototyping. But some players believe that some characters are still fully voiced by AI.
More well-known studios are also using artificial intelligence for voice acting. One of them is Santa Monica Studios, the authors of the God of War series. In 2023, the studio released God of War Ragnarok, which was loved by both players and critics. Even in a project of this scale and pathos, they still used AI to generate NPC voices.
Interestingly, this was done by the Ukrainian company Respeecher. She also managed to work with such companies as iBlumhouse, Lucasfilm, Deezer, CD Projekt on Cyberpunk 2077: Phantom Liberty and other well-known names from the entertainment industry.
Not very successful examples of AI use
It should be noted here that Squanch Games and Santa Monica Studios were still quite careful in using AI-generated content. Most players may not even notice that the game has something of generative artificial intelligence. But this is not always the case. Sometimes developers depend on AI to a greater extent.
Because of this, Embark Studios found itself at the center of a small scandal, which at the end of 2023 released the multiplayer shooter The Finals. The shooter from the former developers of the Battlefield series features the style of a TV show and the ability to destroy the environment, which many players miss. However, some users have noticed that The Finals uses AI voice acting very often.
Representatives of Embark Studios admitted that they use a combination of voice acting from live actors and AI content in the game. It all depends on the context of the situation, because sometimes AI voice acting can be implemented in the game faster and easier, relieving voice actors.
“Even when we use text-to-speech technology, all the lines are always based on real people. <...> We have never sought to create games without live actors, technology has shown us new ways to collaborate,” explain representatives of Embark Studios.
However, in The Finals, the vast majority of content is created by the developers themselves. However, the capabilities of generative artificial intelligence have reached the point where almost the entire game can be created using various AIs. And here we cannot think of a single case when someone would have liked it.
On Steam, you can find the game called This Girl Does Not Exist. It costs 124 UAH, which is a relatively small amount of money. It is a very simple puzzle in which you have to put together the faces of different girls. The nuance is that absolutely everything in this game, from characters to music, is created with the help of AI.
This Girl Does Not Exist cannot be called a very popular (or high-quality, for that matter) game. But it was the first project on Steam to be created entirely with the help of generative artificial intelligence. Players greeted the game with a flurry of criticism, mostly related to the fact that AI was used so actively in game development. People were worried that these technologies would put them out of work, and the game itself has “crap vibes.”
“If I would do a whole AI game again. I would maybe try to tell people only after they finish the game and see what would the reactions be, as I think with This Girl, maybe people had prejudices up front,” says one of the developers of This Girl Does Not Exist.
Cute Pen Games, the studio that created This Girl Does Not Exist, continued to use generative AI in its games, albeit not as actively. The studio shifted the focus of its projects to erotic and sexual content, which is in great demand on Steam.
By the way, there is also a Ukrainian project on Steam, whose developers claim that almost everything in the game, from code, voices, images, to puzzles that the player has to solve, is created by AI. The game in question is The Riddler, created by one of the members of the anonymous collective Bunker 22, the authors of Putinist Slayer.
Problems of definition
On Steam, there are tags “Artificial Intelligence” and “Procedural Generation” that help to identify games with active use of generative AI. The problem is that not all developers label AI content, and these tags can include other games that are related to the topic from other angles.
For example, here you can find Five Nights at Freddy’s: Security Breach (because the local animatronics are also technically “artificial intelligence”) and The Binding of Isaac: Rebirth (because the levels are generated procedurally). Steam as a platform has also not introduced any regulation or categorization for AI games.
And this is a big problem because the active use of AI in the gaming industry can lead to very bad consequences. The most obvious one is the loss of jobs. Why hire an artist or animator when AI can do everything for them? AI does not need to pay bonuses and sick leave, provide equipment and a workplace. From the point of view of corporate spending, technology can help optimize costs, albeit at the cost of jobs for real professionals.
Massive layoffs in the gaming industry over the past two years have only increased fears of rejection of generative artificial intelligence among gaming industry professionals. And this is at a time when top management of companies is actively considering AI capabilities. Andrew Wilson, CEO of Electronic Arts, drew attention to them.
“The development cycle is very iterative, and incredibly time consuming, which is why games are now taking six and seven years to build, as they’ve gotten bigger and deeper. So the first thing for us is how do we make that more efficient? <…> About 60% of all of our development processes have high feasibility to be positively impacted by generative AI,” says Andrew Wilson.
The use of generative artificial intelligence is also associated with uncertain legal nuances. After all, it often happens that it is not known for sure on which database the AI was trained. As a result, it may include materials from authors who did not consent to the use of their work for such purposes. That is why artists do not like artificial intelligence.
That is, in a sense, generative artificial intelligence only “digests” everything it is fed and tries to produce the result that best matches the database. Here we are already entering the territory of discussing general art, because great artists also regularly copied each other. Conflicts around AI are complicated by moral and ethical issues.
The lack of competent and comprehensive legal regulation also does not help the situation. However, things are slowly getting better here. For example, the European Parliament is already working on a program to limit the risks associated with the use of AI.
Real benefits from AI
But does this mean that generative AI in game development is a definite evil? After all, neither players nor game creators like it. However, generative AI has already proved its usefulness for developers.
Horizon Zero Dawn and Horizon Forbidden West are known for their large and very beautiful open worlds, which can take tens of hours of real time to explore. The developers from Guerrilla Games could have created these games much longer, but they were helped by procedural generation.
Not all locations and objects in these games are made by hand. The developers have created a special algorithm that fills the area with vegetation, stones, and other objects. Everything is optimized to make the game world look alive. The Decima game engine uses the power of the GPU to create the environment around the main character Eloy. Meanwhile, developers can do other tasks.
This example is also important from the point of view of the legal justification of generative AI. Guerrilla Games did not use anybody else’s content. They created their AI for the engine entirely on their own, based on their own rules and assets. The result is a very useful tool, and the feedback from players and critics about the beauty of the Horizon universe confirms that it was all worthwhile.
Developers of the Unity engine are now making a big bet on AI. They are gradually introducing new features of using generative artificial intelligence in their development. The authors of Unity see the following opportunities in AI:
- improving NPC behavior;
- generating two-dimensional and three-dimensional assemblies;
- creating error notifications and participating in QA testing;
- writing program code;
- improving gameplay mechanics;
- development tips for developers themselves.
There are also AI plugins for Unreal Engine. They allow you to add individual objects and create entire game mechanics and systems.
Therefore, generative artificial intelligence will definitely not disappear. This also applies to game development. The main thing here is to perceive AI as a useful tool, not a replacement for developers. Because the latter will not work out well for anyone.
Loading comments …