It’s starting to feel like every area of the tech world is dipping its toes into artificial intelligence (AI). Industry leaders truly believe that the integration of AI capabilities will transform the way we interact with our devices and the world around us, and they’re making sure we know about it.
From phones to washing machines, we’ve been hard-pushed to find any upcoming devices that don’t have some AI-powered features as a key selling point.
It’s interesting, then, that the gaming industry has been relatively quiet when it comes to AI developments, because its potential in this field is just as great. Perhaps it’s because AI isn’t a particularly new thing in gaming. Machine learning, context-sensitive behaviour, neural networks, natural language parsing – all of these techniques have been pushing games forward for years now. Black and White may have been released in 2001, but it’s still one of the best examples of machine learning in a game.
However, it could also be because in recent years artificial intelligence is becoming rapidly more, well, artificially intelligent. With more complex machine learning systems evolving, we think things could greatly change for games over the next few years as developers embrace them.
Improved visuals
One of the most immediately noticeable things AI could improve in gaming is its visual quality. With deep learning and the ever-growing amounts of data at our disposal we could see game environments and game characters become more realistic and natural than ever before.
When TechRadar spoke to Unreal Engine developer Tim Sweeney he told us that he believes advancements in AI have the potential to greatly improve the way game characters move and express themselves, and create something that’s much more realistic and natural.
“I think there’s a lot that can be gained from incorporating more advanced and modern forms of AI into our game processes,“ Sweeney said. “Especially in the areas where we can use a huge amount of data to have algorithms figure out how humans act in the real world and extrapolate that in the game world.”
By scanning thousands of hours of footage of people interacting and mapping it onto game characters, he added, “we could see more human emotion” beyond what we can achieve through technologies such as motion capture.
Even the way characters move could become more believable. “Most games have a character animations driven by several hundred canned animations that were motion captured and they use a very old school algorithm to piece these chunks together,“ Sweeney said.
"What you’d really like to do is scan thousands of hours of people walking around in the real world and use that to figure out how to get a game character from one location to another. Because it’s not always turn, walk in a straight line and stop, right?
"There’s a lot of interesting possibilities to it that can be discovered through deep learning applied to a huge amount of data. That will really make games improve hugely in terms of visual quality over the next 10 years in some unexpected ways.”
However, there are also ways in which AI could work behind the scenes to improve our games without such immediately obvious results.
Real-world ramifications
Right now we probably most associate AI with the voice assistants that are being integrated into our home speakers and smartphones. Bixby, Alexa, Google Assistant, Cortana, Siri – you’ve probably had a stunted and uncomfortable conversation with at least one of them.
Though they’re very much a work in progress, it’s worth noting that these assistants are already changing the way we play games.
These assistants support a wide range of interactive games, from BBC dramas that play like a Telltale narrative adventure to interactive trivia games for the whole family. In addition to hosting their own games we’ve actually recently seen these assistants being worked into existing console games.
Destiny 2, for example, now has its own Alexa app through which players can more easily access certain game features without having to repeatedly go through aggravating menus.
On a more immediate level, then, AI voice assistants could greatly change the user interface and user experience in games if voice assistant apps became an industry standard.
Further down the line we could perhaps see the way these assistants are integrated with our games become more evolved in a way that would bring our virtual and real worlds closer together.
The CEO of Electronic Arts, Andrew Wilson, has previously posited, in an interview with The Verge that in the future our lives will become entwined with video games as they move from being “a discrete experience to an indiscrete experience“. Wilson gave a host of examples of ways in which our real-life actions could impact the ways in which we play games.
“From the minute I get up in the morning, everything I do has an impact on my gaming life, both discrete and indiscrete,“ he said. “The amount of eggs I have in my internet-enabled fridge might mean my Sims are better off in my game. That length of distance I drive in my Tesla on the way to work might mean that I get more juice in Need for Speed. If I go to soccer practice in the afternoon, by virtue of internet-enabled soccer boots, that might give me juice or new cards in my FIFA product.”
With a well-integrated AI system this is something that has potential.
One day, when the AI personality leaves its speaker and the augmented reality technology reaches its peak, we could end up with digital storytelling and gaming experiences reminiscent of Westworld – as potentially terrifying as that sounds.
Personalized play
The concept of personalized service through AI could also change our games without necessarily bringing them more directly into our real lives. There’s more data about individual players out there than ever, and improved AI means more ways to process it.
If AI was able to collect data on players and understand their preferred style of play, we could end up with game worlds that are more reactive to individual players in-game.
With the help of AI we could see MMORPGs that push players in the direction of quests and other players which compliment their style, more intelligent NPCs who feel like they’re naturally conversing with your characters, more challenging enemies who learn your combat patterns.
We just need to look at Alpha Go to know that Deep Learning AI can beat us at our own games. Repeatedly and without mercy.
What does need to be considered, however, is that when AI is used in games in this manner there could be a point where it stops becoming fun; when an AI behaves through learning rather than programming there’s an element of unpredictability that would have to be monitored. Should an AI go haywire or make it impossible for a player to win, the fun and empowerment that comes from gaming could be lost.
Design tools
It’s worth, then, considering the role AI could play in overall game design rather than just in changing game content – to work for game designers and impact their work rather than to directly impact players. With enough developments we could one day see this AI and data collection work together to empower designers to make the best possible systems and decisions for their creations.
Running hundreds of simulations and scenarios through an AI system has the potential to make fine-tuning during the development process easier, to find different strategies and test a game as many different kinds of players.
Whether it’s changing the way we interact with our games, the content within them, or the way they’re designed, it’s undeniable that AI will continue to revolutionize the gaming industry – perhaps at a faster rate than it ever has before.
- TechRadar's AI Week is brought to you in association with Honor.
