Artificial Intelligence (AI) is an intelligence manifested by a machine, a program, or an artificially created consciousness. These include automated machines that perform tasks for human intervention, elements that respond to our impact in games, and self-learning, communicating AI’s.
Artificial intelligence used to be just a product of the imagination, many books and films told that man-made machines and programs think independently. Today, the development of AI technology is a major branch of science that deals with intelligent behaviour, machine learning, and machine adaptation. Today, machines are capable of controlling and scheduling systems, performing diagnostic tasks, answering questions, recognizing speech, face, fingerprint, and more. Artificial intelligence systems are widely used today in economics and medicine, in the military, in many computer programs and video games.
In video games, AI is used to generate intelligent behaviour, especially for non-player characters (NPCs). Artificial intelligence has been an integral part of video games since the early 1950s. The role of AI in games has greatly expanded since its introduction. Modern games often use existing techniques in the field of artificial intelligence, such as routing and decision-making mechanisms, to control the activities of NPCs.
Some argue that video game AI has revolutionized the way people interact with all forms of technology, although many researchers are sceptical, especially of the idea that such technologies fit the notion of “intelligence” commonly used in the cognitive sciences.
Researchers at AI have argued that video game AI is not true intelligence, but a simple computer program that uses sorting and matching algorithms to create the illusion of intelligent behaviour. Because the presence of NPCs focuses on the emergence of intelligence and gameplay within environmental constraints, its approach is very different from traditional AI.
Old games like chess, Pacman, or any machine that can be played against fighting, strategy, or action games used AI. Today, there are several varieties of artificial intelligence that appear in video games, obviously depending on what role they play in the game.
Another gaming-related area that is affected by AI is the online casino world. In fact, these two terms have met 20 years ago when BM’s Deep Blue computer defeated renowned chess master Garry Kasparov 3.5–2.5 in a six-game match. After that, the AI was improved even further and arrived in the casino industry as well. Now you can find implemented AI systems in slots, roulette, blackjack games, and so on. The AI is used here to improve our user experience. But the AI was tested as a player as well, in the form of Libratus which is a form of AI. In January 2017 Carnegie Mellon’s Libratus AI has won thousands of hands and even defeated top-rated players.
But let’s talk about action games. You’ve probably seen how silly NPCs sometimes behave. Well, that shouldn’t be the case, because technically AI can be a lot smarter in a game than man. The behaviour of individual NPCs, as I wrote at the beginning, can be determined manually, and it is true that these handwritten action trees can often make funny mistakes. If set incorrectly, the figures will get caught in the tree, make repetitive movements, disappear from the track, or blend into the environment, and even produce unexpected behaviour. Such is the case when you drive regularly in the GTA and a passerby jumps from the sidewalk in front of your car.
If, say, in a fighting or stealth game, the opponent really had the abilities of a real soldier, the player would always fall. This is why they form “weaker” NPCs for players, depending on the difficulty level of the game, of course. If the AI is too smart, too human, or just has the actual abilities of a superhero, the player’s chances are reduced, which goes to the detriment of enjoyment and the experience of success. Cheat programs have also been created against very strong AIs.
Surely you are also familiar with games that are the hardest way to play for just a few people. In one part of the Otherworld channel’s Otherside series, Phelan discusses why there was an impressive team play on the part of the opponent in Star Wars: The Force Unleashed (Sith Master mode).
He also talks about what mechanics NPCs move based on, why they don’t go beyond a point when they see it, and what they hear. Their set values determine, say, how much a soldier hears on the field while spectacularly stabbing his partner behind him. If the adversary always spits on us, it is a less human, more robotic behaviour, so refining that is key. The examples in the video also show that in many cases, AI seems to be a bug in games if the developer has not set its parameters properly. Of course, who has what is easy or difficult is subjective. This is exactly the sport, the essence of e-sports.
I don’t want to go into exactly how video game AI development works, as there’s not a lifetime to do it, but maybe I was able to show how its operation differs from the practical area of AI development. But let’s look at some specific games where AI works particularly well.
A first-person shooter where the player finds himself facing supernatural phenomena and strange soldiers. Here, AI works with a context-sensing mechanics that was first used in mainstream games. Opponents are able to use the environment wisely, hide behind tables, seek cover, open doors, breakthrough windows, depending on the situation. Teamwork is also noticeable, and even soldiers manoeuvre to the side, use a firing fire, and so on.
STALKER series (2007-)
In this shooting horror game, you have to face strange creatures, soldiers and mercenaries. Various enemies (if the difficulty level is set to the highest) use combat tactics such as healing wounded allies, bypassing the player. The AI uses completely different tactics in the same area each time you try in the game (even after reloading in the event of death).
Far Cry 2 (2008)
This Far Cry series may not need to be presented to anyone. AI works in a game based on behaviour and decision, meaning it is unpredictable what your opponents will do. They respond to sounds, environmental changes such as fire or nearby explosions. The player is able to distract the AI, meaning the player can turn the confusion to his advantage. The AI senses the player’s proximity and can spot it, even if it can’t see it. If an opponent is injured, he can shout and alert other teams, who can call for further reinforcements. The range of possibilities is finite and can be used after a long game, but artificial intelligence has really worked well.
It’s rare to come across as intelligent and tense AI as we’ve experienced in Isolation. Whether it’s the robots or the Alien itself, we’ll need all of our tactical senses so that the game doesn’t end soon