With AI making the headlines in 2023, you’d be forgiven for thinking that it materialised — out of nowhere — into tech and gaming just this year. The truth is, however, that AI has permeated the gaming industry and shaped video games since the very beginning. Until multiplayer games emerged from the neon-drenched arcades of the 80s, gamers played against AI opponents, like the paddle in the 1972 classic Pong. The capabilities of AI have snowballed since then, with recent developments ushering in the era of generative AI, which is set to fundamentally change gaming as an industry and a medium.
Gaming has long been an early adopter of new technologies. With every release, gamers expect better graphics, bigger worlds, and more interactive non-player characters (NPCs), and that’s just the start. In an increasingly hyper-competitive business landscape, with more public scrutiny over game releases than ever, game developers can’t rest on their laurels – they need to be ambitious, constantly iterate, improve, and offer more to players. This leads to a problem: with time and money being finite resources, how can games get exponentially better?
Cyberpunk 2077 – which took eight years and €260 million to develop – was met with raised eyebrows upon release in 2020, being perceived as unpolished and not completely bug-free. It was inconsistent with developer CD Projekt’s lofty ambitions and promises. While the developer did well in investing resources to iron out the issues and improve the game’s stability and quality to the expected level, the solution must be more than just putting additional pressure and demand on already stretched game development teams.
The answer may instead lie in giving AI more responsibility.
Gen AI, year 0: What AI looks like now
Recent breakthroughs mean AI’s capabilities – and subsequently the value it provides – are at an all-time high. For example, Natural Language Processing (NLP) improvements have significantly enhanced AI’s ability to understand and generate language — think ChatGPT. New generative models mean AI can create images, videos, and audio — take Midjourney, for example. Reinforcement learning means AI can learn by playing against itself. This is how AlphaGo trained itself to master the board game Go, a victory long thought outside of AI’s capabilities given the game’s close-to-infinite possibilities.
These advancements have not occurred in a vacuum. They result from broader macro trends: improvements in hardware (graphics cards) and software (cloud/distributed computing), the creation of large datasets (like ImageNet), and an increase in specialised labour. As this continues, the abilities of the AI tools that we use daily will grow and grow, helping make games exponentially better – the industry’s goal.
AI will first impact traditional processes like game creation, distribution, and operation. Then, we will witness the rise of AI-first games and the furthering of human-machine interfaces with AI.
Simpler, Better, Faster, Stronger: How AI Changes Game Production
Game creation requires a host of different kinds of content. If these can be automatically generated then production times will be radically sped up. We’ll see automation of the visual elements of games – including 2D assets (e.g., Scenario) and interfaces, 3D assets (e.g., Sloyd), character animations (e.g., Kinetix), and fluid simulation (e.g., Zibra) — as well as game audio, text — and even the code. Considered chores by game developers, code quality assurance and game localisation will progressively be automated. While not immediately apparent to gamers, shorter development times and more ambitious games – from both small and large studios – will become the norm.
Generative models will also simplify game distribution, as teams can use them to develop and roll out robust marketing plans and create the required text and visuals. Clips automatically generated from streams, utilising AI tools like Powder, will unlock organic, creator-led game discovery at scale.
Some of today’s most popular games are evergreen: instead of providing a set amount of gameplay, as with the old model, games are refreshed with content regularly – something the industry calls ‘LiveOps’ (Live Operations). AI can support LiveOps by ensuring the new characters released into the game don’t break its balance – and also by making sure games are a safe space to play in with AI moderation (e.g., Bodyguard).
Revolutionising Play: Towards AI-First Games and New Human-Machine Interfaces
Excitingly, we’ll see new types of games in the medium term. The culmination of AI in gaming will materialise when games adapt themselves to the gamer — visually and from a storytelling perspective (e.g., Scriptic). We’re already seeing the first signs of this content in the video and music industries, with algorithms recommending hyper-personalised digital content for users. In the future, games’ storytelling will be adaptive, as users interact with intelligent NPCs, essentially AI agents playing an active role in your progression. Currently limited to a large set of scripted responses, NPCs will hold actual conversations, offer opinions, and react authentically to players’ actions.
Besides unlocking new types of games, AI promises entirely new ways of interaction, transforming how we experience the medium. Voice – and even thoughts! – will progressively develop as credible alternatives to touch screens, controllers, the mouse, and the keyboard. Players will talk to their device – or even think about how they’d like to play the game like Neuralink’s Monkey MindPong does.
And it’s not just about controlling what happens in the game, but also about being able to create content in it freely. Researchers have demonstrated convincing thoughts-to-images technology. Is the ultimate evolution of speech-to-text thoughts-to-3D-worlds creation?
AI is making ultra-fast progress that will profoundly impact the gaming ecosystem, and for all the progress made, 2023 is just the start. We’ll get more expansive, higher-quality games shipping faster in the short term. In the medium term, expect AI-first games and new ways to interact with them. AI will massively contribute to a healthier and more successful gaming industry. We’ve come a long way from Pong, and are set to go much further.
Lead image: tohamina