The application of artificial intelligence in gaming is all about generating more reactive, adaptable, and demanding experiences. In spite of academics' contentions that gaming AI is not really artificial intelligence, the buzz around the technology has been steadily rising for quite a while now.
Pathfinding and finite automation have been the mainstays of AI in video games up until now, but this is about to change. AI-controlled NPCs use pathfinding to determine where they can and cannot travel.
This is still relevant, regardless of whether you're playing a classic old-school Mario-style game or a big and elaborate RPG. Virtual residents of a community should not be able to walk through barriers or get trapped in the earth. In addition, developers don't want them to remain still at any given time; they are supposed to move like real people.
Finite automation, on the other hand, enables the AI to adjust its behavior depending on pre-determined parameters. To illustrate this, we can take nearly any game incorporating a hostile environment and lots of enemies.
At the moment the player’s character passes through a sightline of an enemy, AI signals that the character has been detected. Depending on the settings, this NPC can signal for reinforcement or “decide” to solely attack the character.
This technique is used a lot in stealth and action games. Also, you must remain out of sight and unobserved long enough for the security personnel to abandon their pursuit and return to their original places.
These actions can be called “Hunting” and “Patrolling,” and they are basically two separate modes of operation.
A lot of AI programming today relies on these two fundamental principles even if they look simple. For example, let’s take simulation games. In probably the most famous one, SIMS, NPCs try to behave like real human beings. If they want to eat, they will start searching for food. After a long day, they go to bed when they are worn out.
Sounds pretty human-like, right? However, AI doesn’t know about being “hungry.” It just switches the modes from “idle” to “time to find some food.” Then, the program establishes the route, pointing NPCs in the right direction.
So, what is the development route of AI? With the vast quantity of pathfinding and states that developers now have at their disposal, this has already undergone a significant transformation. After all, Mario and Red Dead Redemption are quite different games.
Today's self-driving vehicle and face-recognition AI technologies, on the other hand, will permanently alter gaming's AI landscape. Pathfinding may no longer be limited to directing an AI in the direction it should go.
It's hard to imagine what type of narratives we might tell in video games if we could give non-player characters (NPCs) real feelings, memories, desires, and aspirations, as well as a level of intellect equal to our own.
Experiments at the Expressive Intelligence Studio have brought data scientists a step closer to creating AI that can express actual emotions. We may soon see this AI in video games if it is successful.