The AI trends've been becoming a hot topic around the world. In this post, I'm going to share my thoughts about those trends in the game industry. I referenced some documents and combined my experiences in Hogvalord and Hogvalord: The Ranch projects to bring the best content for you I can. I might make some mistakes on this topic due to my limitations, so your feedback is helpful to me. Don't hesitate to drop your opinion in the comment.
ACADEMIC AI AND GAME AI
In my opinion, we need to separate two schools of AI. One is academic AI, and another is game AI. The current AI trend is academic AI, and its history began in 1950. Turing is considered an adopted father of this AI school. The game AI history started in 1979 along with the Pacman game.
The academic AI goals are understanding the mechanics of the human brain and mental processes. Another goal is to create machines that can perform human tasks. By contrast, the game AI's goal is to make a believable experience for the players and help them escape from their sorrow and stress in reality.
The game AI is simple, small, and works in the simulation environment of a game. The application of game AI is more realistic than academic AI. The scope of the game AI is small and has specific characteristics. Because of that reason, AI researchers leverage them in their research and training agents in academic fields. For example, DeepBlue can play Chess, AlphaGo can play Go, and DeepMind and TorchCraft can play StarCraft.
There's nothing has changed much in academic AI in decades. But the up-rising of ChatGPT and OpenAI in 2023 surprised the whole world. That's the result of heritage from the works of predecessors and marketing. This explosion has had an impact on the gaming industry. Personally, I think it's time the academic AI retributes to the game industry after many years of gaining value from the game AI. I list all the recent AI application trends I can think of in the below section. Some of you might recognize them in the game industry recently.
THE APPLICATION TRENDS
The first trend is content auto-generation in game development.
The advanced searching algorithms and vast data resources created a good foundation for academic AI to step out of its theoretical zone. The easiest trend we all see is AI platforms generate 2D images, 3D models, videos, sounds, music, script, dialogue, and stories for video games. The rumors said those applications would help game developers and publishers save production costs.
There are some examples of this trend in the game engine industry. Unity has recently presented Unity Muse for asset generation. Unreal Engine also has tools such as PCG or MetaHuman.
The second application is to improve the realism when players interact with the game elements.
The NPC conversation, for instance, could be more convincing. People expect this application will create a unique experience for each player. However, we can't overuse AI due to the risk of breaking the player-playing session. With this in mind, let's look at the case when the NPC refuses to give the quest item to the player because it hates the player's communication manner. It sounds unreasonable, but there is a high chance to occur when you don't control your own AI platform. I wonder how big studios solve this issue. Not only does the solution relate to a technical issue, but it also depends on the game design. I get a headache when thinking about this problem.
The third application is making the game character's motion more convincible.
This application will help to boost the quality of 2D and 3D character animations. Animation software can apply AI physics to help with animator's tasks. Or using AI to clean up the motion capture raw data quickly.
Then again, AI also enhances the game characters' motion when they travel through complex terrain in the game environment. Some examples are walking up or down stairs and jumping forward or backward on different-shaped stones. Having said that, this application only works for the complete game. We need to retrain the AI every time we adjust the level design.
The fourth application is about testing automation.
In my opinion, this application is the most helpful for solo developers or small-size teams. I had to repeat testing my game Hogvalord over hours anytime players reported bugs. That process is very tiring and time-consuming. I wished I had the AI for the testing automation to help me with that. But the price of that kind of platform could be the biggest concern for the solo indie developer budget.
The fifth trend could be AI usage for cheating and anti-cheating.
The players can use AI to mimic human-playing style to pass the traditional anti-cheat approaches in MMO games. On the other hand, publishers could leverage AI for anti-cheating to create a fair environment for every player. This trend sounds like an AI war in the virtual world, the battle between yin and yang.
The last application I can think of is modeling a player's behavior.
The game studio or publisher can use AI to analyze the data about players' behavior to enhance the design process of games. It can complement the game production and offer better customer service. However, AI-based game design can make the game soulless. Players could turn their back on the game does not have the game designer's unique mark.
CONCLUSION
From what I see, those AI applications could drive us to a new horizon of possibilities and risks. It can create conflicts with the art of game design when we abuse AI in game development too much. We also need to be careful with the bubble AI and make sure it is not a fashionable technology like NFT or blockchain games. I keep watching it while making my game Hogvalord: The Ranch.
Comments