When cooped up at home due to lockdowns, millions of people have turned to games for handling everything from boredom to stress. Even as other forms of entertainment like the movies take a hit due to the pandemic, gaming looks relatively better off.
Game developers are always on the look for new technologies that will help them create cutting edge games that players fall in love with. Machine learning seems to be the ultimate technology that will change the gaming world as we know it and here is why…
Machine Learning in Game Development
Machine learning (ML) is an exciting subset within the wider subject of Artificial Intelligence. The evolution of ML within AI research circles over the last decade has been phenomenal. From its very early days, gaming has played an active role in ML and AI research.
There are two main ways in which game development and AI research are linked. For starters, video games rely a lot on decision making – the software opponents have to be “smart” to present an enjoyable challenge to the player.
And in recent years, improved graphics hardware (GPU) has enabled game developers to create more and more detailed and realistic game worlds. Creating a mind-blowing world like the Wild West in Red Dead Redemption 2 requires dozens of designers and hundreds of hours of work.
But ML can play a key role in reducing the burden on human game designers. This can involve taking control of in-game effects like weather, faster rendering of visuals when you get close to objects, or even create new worlds on the fly.
Making the Game AI “Smarter”
We have not even begun to touch upon the impact that artificial intelligence will have on companies and the economy in the coming decade. We are still years away from creating AI that has a fraction of the complex intelligence of a human. Video games, particularly those that involve complex decisions (strategy games, tactics, etc) always suffer due to the inability of the game AI to present any kind of challenge to players.
Opponents and NPCs (non-playable characters) in games these days rely on scripts – making them entirely predictable and boring in the long run. The advent of smarter NPCs that can be taught to improve their skills and behavior would revolutionize gaming as we know it.
On the one hand, players will no longer have to head straight to online-multiplayer for a genuine challenge. Playing against other humans, while fun, is not everyone’s cup of tea – connectivity issues, toxic behavior, online gaming sure does have its share of niggles.
With smart AI based on machine learning, single-player games will become more compelling than ever before. And even in online multiplayer, you will be guaranteed a decent challenge even in the absence of other gamers.
Faster Game Development
With machine learning, Valve might even get around to making Half-Life 3 – the most awaited vaporware in gaming history! Jokes aside, game development cycles are insanely long – it took CD Projekt Red over 7 years to get Cyberpunk 2077 to the market, in a buggy mess to boot.
Game developers get a lot of flak for pushing out half-finished products. And a lot of that criticism is fair and justified. But we cannot ignore the level of effort required in creating a modern AAA game. Designing the game world, programming the AI – all these are labor-intensive tasks.
A lot of these require hard coding from the developers – the NPC behavior scripts we mentioned earlier is a great example. In highly complex worlds like RDR2 or Cyberpunk, you can have thousands of unique NPCs – each requiring extensive scripting.
Equipping AI NPCs with machine learning would remove a lot of that workload. Unity, the popular game engine, has had basic ML learning tools since 2017. Then there is the issue of creating virtual worlds – if you want maximum realism (like in RDR2) real designers need to spend long hours on it.
The other route to designing game worlds – procedural generation – uses algorithms to create unique worlds, using images and graphical data provided by the game designers. But procedurally generated game world in the past lacked realism and polish.
But it has steadily improved over the years – No Man’s Sky is a good example of a game that uses procedurally generated worlds to provide infinite replayability. With smarter AI and ML, game developers can soon look at creating procedurally generated worlds that rival AAA game worlds and way faster as well.
More Realistic Graphics
In the last couple of years, Ray Tracing – a technique that replicates the effect of light in video game worlds – was in its primitive stages. The hardware was not easily available to the average gamer, and video game companies did not have much compulsion to develop games with Ray Tracing.
But with the launch of next-gen consoles and cheaper RT-enabled graphics cards in the latest Nvidia 3000 and AMD Big Navi series, Ray Tracing is about to become a regular feature in video games in 2021 and beyond.
It requires complex algorithms to design how artificial light behaves in-game worlds – machine learning will make life a lot easier for video game designers and developers. Since we now have high-end GPUs capable of running ML algorithms with increased efficiency, new games will benefit from realistic lighting, better textures, and faster rendering.
Big Data – Challenge or Opportunity?
The biggest roadblock to developing ML capabilities in gaming is the lack of enough data. AI requires vast troves of relevant data to improve its learning capabilities. Right now, we are still in the early stages of accruing this kind of data in the gaming industry.
But as more games launch with ML-driven technologies, this will change and fast. It just takes a bit of time – with recent advances in software and hardware, we are accelerating towards that inflection point. This is an exciting time to be a gamer!
Caroline is doing her graduation in IT from the University of South California but keens to work as a freelance blogger. She loves to write on the latest information about IoT, technology, and business. She has innovative ideas and shares her experience with her readers.