Gaming’s Inextricable Need for AI

Artificial intelligence has always played an important part in video games. From testing worlds, creating worthy opponents, and checking reliability, AI is one of the industry’s most fundamental and inextricable components. Even as multiplayer gaming takes over increasingly more of the worldwide gaming market, the need for AI is unlikely to change. So how is this used, and what standout examples illustrate the benefits of AI?

In the Beginning

The very first video game implementation of AI came about earlier than most would think, arriving for the computerized game Nim in 1951. A mathematical strategy game, Nim utilized basic algorithms to regularly beat even higher-skilled players, and well before the term video game would enter the global lexicon.

In terms of more modern recognizable games, bigger examples of single-player AI wouldn’t hit the mainstream until 1974 with Taito’s Speed Race. While primitive by today’s standards, the seemingly reactionary behavior of digital opponents fascinated players. Such fascination would only grow through the 70s, as titles like Space Invaders would continue to raise the bar.

The level of intelligence these games displayed continued to improve with technology, and, while there are many examples to draw from, perhaps the most important is illustrated by the FPS bot. Specifically, this term refers to opponents in multiplayer first-person shooters designed to mimic human players.

These bots would take the idea of a computer-controlled opponent and attempt to fuse it as close as they could manage with actual human ability. Rather than operate as a completely separate entity, these bots would fill an empty spot by ostensibly operating with the same limitations as human players.

Quake 1’s Reaper Bot was a famed example of FPS bots in action. Developed by a fan of the game, Steven Polge, in 1996, this bot could learn each level as it played them. It could find the best weapons, figure out how doors worked, and even become a camper if it proved viable to do so.

While this bot could be manipulated into failure, it still proved an experience far more analogous to human players than early developers thought possible. So successful was Polge’s bot, that he would eventually go on to be hired into the gaming industry. Now working for Epic Games, Polge is one of the primary minds behind the AI in the Unreal Engine, one of the best and biggest in the industry.

Artificial Intelligence or Digital Stupidity?

As most forms of artificial intelligence in gaming came from developing opponents, it was inevitable we’d hit a wall. Humans are masters at a great many things, but we’re also creatures of mere biology. To a machine, the best humanity can do would never be enough to win.

In other words, AI in most games running at their full potential would destroy any human player. They could read your inputs, counter perfectly, and in doing so drain the experience of challenge or enjoyment. Be as great a fighter as you want, you’re still not beating AI.

With this in mind, the problem became how AI opponents should be dumbed down. You can’t have an enemy instantly kill you from across the map, that ruins the fun. On the other hand, having an opponent in front of you who refuses to do anything but dodge around like a henchman in a Batman movie ruins an otherwise engaging experience.

This is an issue of balance, where the right opponent needs to appear a threat, while still being reliability beatable. For many teams, this means creating a perfect killing machine, and then inhibiting it until it becomes almost human. Just as much work can go to making something stupid as making it smart, an interesting dichotomy. 

When this works, you have enemies like those in Half-Life or FEAR, both of which were famed for their flanking and covering tactics. This made encounters a constant give and take, where an enemy’s threat made defeating them that much more thrilling. On the opposite side of the spectrum are games like Aliens: Colonial Marines, where a simple typo turned terrifying xenomorphs into bumbling oafs.

Safety Testing

Outside of offering opponents, one of the most consistently useful aspects of artificial intelligence is reliability testing. As software glitches and bugs are an inevitable part of the programming landscape, extensive testing is required to achieve the best possible product.

The problem with this is that testing can be expensive and time-consuming. To overcome this obstacle, some games have found success in implementing AI testing procedures to streamline this cumbersome process.

This form of testing is commonly used in online casino games. Here, titles like Buffalo and 5 Dragons can be played for free or for money, therefore, these games need to be safe and reliable for players. Testing these by human hand is untenable, so AI is used instead to run thousands of tests in a fraction of the time. This way, mapping and addressing potential bugs before a game goes live becomes infinitely easier.

Another standard use for safety testing is in website flooding analysis. Dedicated denial of service (DDOS) attacks can be emulated by AI software, which allows websites to gauge their protective systems. If your website keeps working, great, if not, then you’ll at least be aware and able to investigate.

An Industry Built on AI

AI systems form the backbone of most interactive entertainment experiences, but it can be difficult to appreciate them. Working properly, these can convince you that the character you just killed was being controlled by a person, or ensure your game never crashes. If not directly brought to your attention, it’s easy to overlook these elements. Even though we might not realize it, AI is always there, waiting in the wings, and covering us where we fall short.