Machine learning techniques are being applied to all sorts of technologies, and it's only natural to wonder how those could be used to empower game development.
That's exactly what Google is researching for developers through its Project Chimera. A team of engineers and developers have been looking into the potential applications of generative adversarial networks (GANs).
Speaking with MCVUK magazine (April 2020, issue 956), Erin Hoffman-John, head of creative for Stadia research and development, explained that machine learning could allow small development teams to create even World of Warcraft-sized games. To begin with, content creation can be made much simpler as machine learning trains on a set of reference images and then produces completely new designs based on that style.
We’re taking on the risk that developers don’t want to. We’ve been talking externally to developers and asking them, what are the things that you’ve always wanted to do but have not been able to do? What are the things that you’ve had to cut out of your games because you haven’t been able to do them fast enough, or you just haven’t had the processing power?
What if a team of 14 people could make a game the scale of World of Warcraft? That’s an absurd goal, right? The thing about games like WoW is that they rely on a lot of heavy, repetitive content creation. The artists and the writers are doing a lot of essentially duplicate work, that’s where a lot of the investment goes. If you look at the amount of money that is spent making a game like World Warcraft, it’s like 70% content and 30% or less code, even though it’s a tremendous amount of code, it’s way more on the content side.
This isn't unlike what we've seen recently with NVIDIA's StyleGAN machine learning, where a generative adversarial network was used to recreate new manga designs based on the works of Osamu Tezuka.
Google's Project Chimera also aims to make balancing much easier for game developers. This is once again something that usually would be very complex to do thoroughly for smaller teams, but machine learning (more specifically reinforcement learning) can go a long way to fix that, according to Erin Hoffman-John.
[...] by playing the game millions of times with reinforcement learning agents that we’ve trained on the rules of the game, that lets us test the balance very, very quickly. So even a small developer who might not have access to hundreds of people to playtest their game could have access to this reinforcement learning tool that will optimise the play of the game. It can learn the game by itself without being scripted and then tell you where the problems are in the balancing. It lets you test your theories of the design against what’s actually happening in real time.
Again, there are already examples of reinforcement learning. For instance, DeepMind's AlphaStar AI is already capable of beating 99.8% of Starcraft II human players. However, the goal here would be to help balance the game rather than just use machine learning to beat humans. Additionally, for both techniques discussed, their effectiveness could very well vary between game types.
Still, it would be a significant advancement if such large games could be successfully made even by those studios that don't have thousands of developers at their disposal. Machine learning is sure to play an increasingly big role in gaming, and we'll keep you up to date on its applications.