In nearly every industry, AI is making a huge splash. AI is helping both marketing and customer support with chatbots assisting customers, as well as helping analyze massive amounts of data, finding new cures and solutions to diseases, helping increase data security, and finding new planets and solar systems in space.
The video game industry is no exception to all this AI innovation. As developers are constantly looking for ways to build new and exciting games, AI is going to play a significant role in how games are built going forward. This could include acting as a game tester, being a key feature in the game, controlling the enemies, analysing how people play, competing against human players, and even building their own games from the ground up.
The Difference Between Real AI and Game “AI”
Many gamers and game developers have been using the word AI to describe how characters and the game acts to a player’s actions. Using this term isn’t really accurate though, as the programming in those games isn’t a true learning AI. Instead, it’s a list of programming reactions on how the game or NPCs should react to different player actions.
For example, if you’re playing Super Mario Bros, the “AI” of the goombas is to move left on the screen unless they bump into something, which then switches their direction. No matter what, that’s what they do. Even if a dozen other goombas fall off a cliff in their path, the 13th goomba won’t learn from that and change directions; it will follow the others off the path.
Modern programming is much more complex than that, but the important part of understanding is that the system doesn’t learn, it’s programmed to act in specific ways. The responses might change from time to time because of setting programming, but they won’t evolve.
Real AI does learn and grow, changing how it acts in new scenarios. In the goomba scenario, after the first one went over the cliff, the AI would try different solutions to avoid the death. That could include changing directions, jumping, or whatever other tools it has.
AI Playing Games and Testing Them
AIs have been playing and mastering games for several years now. AI has beaten world-class professionals in Poker, Go, Chess, Jeopardy, and DOTA 2. It’s clear that AIs are capable of playing games, finding the best and quickest solutions to beating them.
This can mean game developers can test out their games with AI to find bugs, glitches, and unintended ways of playing. They can also have AI at different levels of experience with gaming to see how a completely blank slate plays their game versus somebody with more experience.
This style of testing could be useful for earlier stages of development, but human testers will still be necessary in some capacity. AI, though, could accomplish things humans can’t, like lightning fast inputs to test both the limitations of the software/hardware and seeing different ways it can break the game.
Being a Key Feature in Games
True AI as a game mechanic is probably still a few years off but could lead to very exciting possibilities in how we play games. Realistic war strategy games could have AI trained to mimic the tactics of a variety of historical figures, from iconic Chinese generals to modern-day examples of military strategy. Not just a one time fight against them that do what they did, but actually learn tactics they would use and adapt their strategies in response to what the player does. This could even create intense rivalries between an AI opponent in the game and the player. Imagine playing Pokemon where the rival adapts not just their Pokemon, but also their tactics with them to counter your last strategy.
AI could play a role, not just by controlling the enemies, but also the world as a whole. AIs not only improve themselves, but they can also analyse human behaviour as a whole and on an individual level to get a good understanding of each player’s strengths and weaknesses. Then, they can adapt the game to provide a better experience. A player might find areas they are strong in to be harder, while things they are weak at easier to handle. This could help smooth out frustrations with games, making the experiences more enjoyable.
Building Games from the Ground Up
Already, AI is capable of creating whole games on their own. They might be a bit primitive now, consisting of side-scrollers and basic designs, but they will improve. Sometime in the future, it might even be possible for an AI to make a high-quality game on its own, including quality gameplay and story.
Already, AIs are capable of writing stories and articles, and with enough training will eventually be indistinguishable from human writing. Combine that with the ability to design levels or even entire worlds from nothing, and AI will play a big role in building games.
It’s not just making games that AI will influence, but also the tools used to build games. AI is already impacting the world of IT, by helping fill skill gaps and taking on tedious tasks. With AI, the software used to build games can become easier to use and capable of doing more. Better tools give game developers the more creative freedom to achieve their vision, meaning better games for us to play.
AI is clearly going to be part of our world, both virtual and physical. We’re still several years off from high-quality games made entirely by AI, but it is possible. In turn, developers need to account for the impact AI can have on their games now, both to create new worlds and to become more efficient at making them.