Today, we often associate AI with constant learning adaption. In video games, AI often follows a set of rules to mimic the idea of intelligence. To understand how smart some of these ideas are, we first have to look into the past, and see how it all began.
The Beginning (1980s - 1990s)
Before we delve into the very beginning of video game AI, we first have to differentiate between our definition of AI and video game AI. While our vision of AI is often seen as adaptive and ever-learning, video game AI is usually just complex code designed to mimic intelligence.
AI began to see usage in the 1980s, often in arcade games such as Pac-Man and Space Invaders. Though basic, they were revolutionary for the time due to their random movements that kept many on their toes. These movements often ran on simple math operations.
In the 1990s, AI became more advanced with developments of NPCs (Non-playable Characters) and rule-based systems in RPGs (Role Playing Games). Games like Baldur’s Gate and Ultima gave the illusion of intelligent NPCs by simulating independent decision-making. These developments, however, would be the building blocks for even more advanced AI.
Big Leaps (2000s- 2010s)
In 2009, Scribblenauts was released, an extremely intelligent game that was far ahead of its time and mirrored many aspects of modern AI today. Scribblenauts included a dictionary that recognizes 22,702 words, and through a limited form of AI, it was able to recreate them onscreen using simple graphics. Additionally, the user is allowed to combine multiple objects through this AI.
Also released in 2009, Left 4 Dead was released, and with it, a new system that allowed for a unique play through for each player known as the director. In short, the director takes in the situation of the player and adjusts the difficulty accordingly. If a team low on health, the director would accordingly place more ammo or health packs around the map to help that team. One unique aspect, the director is also able to change the music for each situation, allowing for tense moments at certain points in the game.
One more game that is often regarded as revolutionary is F.E.A.R due to its technique of automated planning. This allowed for an AI to plan out a strategy at an extremely high level, with every action being done due to a precondition. This, in tandem with other AI, makes it seem like each enemy is working together, even though they don't even know that the other one exists.
Modern Day (2020s - Now)
Today, AI is more advanced than ever, with hyper-intelligent AI that can be adjusted and is often used by many professionals to hone their skills. Such examples would include League of Legends with its AI bots and Iracing with its highly customizable AI difficulty.
Additionally, further developments allow for more immersion, allowing people to be better connected to a video game. This is the case especially with Red Dead Redemption 2: small details such as the NPCs commenting on your outfit or doing specific tasks every day allows for better world building.
Finally, AI is slowly beginning to have an effect on our gaming experiences, mainly with graphics settings. Technologies, such as Nvidia’s DLSS, which allow AI to generate more frames and upscales images, leading to higher image quality and a better gaming experience.
Conclusion
AI has come a long way ever since its initial introduction from the arcade machines and low poly graphics (using smaller polygons that result into a blockier graphics style) from the 90s. With video game AI getting smarter every day, as well as actual AI getting involved, we have to wonder how far video game AI will go.
Sources
