Where might AI take gamedev next?

In less than a year, AI has gone from being something most people don’t think much about to the subject of near-daily headlines. Generative AI is already revolutionizing creativity, and it will lead to a giant step forward in video games, too. In fact, our latest closed beta releases are taking steps toward that reality already.

AI brings the promise of making real-time creation accessible to more people, simplifying some development tasks to let you focus on being creative while helping you to achieve more with less. And generative AI is just the latest in a long line of technological breakthroughs that have revolutionized video games.

In 1952, a research product called OXO became the world’s first video game: tic-tac-toe programmed on a massive mainframe computer the size of a room. Atari engineers used an understanding of transistor-transistor logic (TTL) chips to create Pong in 1972, and its simplicity enabled domestic distribution, making it the first commercially successful video game. Space Invaders integrated a microprocessor and a multi-chip barrel shifter circuit into the first-ever fixed shooter, giving the 1978 classic incredibly smooth animations that set the table for this new genre.

The introduction of PCs like Apple II and C64 introduced a broad spectrum of new possibilities – the modular setup with adaptable gravity and physics in games like Pinball Construction Set; the fluid rotoscoped animation of the first Prince of Persia adventure; and the widely ported Lemmings, which saw the game’s simple puzzles released on pretty much every gaming system out there at the time. Consoles really came into their own with 8-bit home systems, which offered unprecedented control, the 16-bit Sega and games like Sonic the Hedgehog, and the console-agnostic sports hits of Electronic Arts, which harnessed that control and realism to bring players inside their favorite sports.

DVD consoles increased visual and speed bandwidth again in games like Road Rash and Need for Speed. The advent and broad adoption of the internet made MMOs possible, bringing gamers from around the world together to play things like Ultima and World of Warcraft. The 32-bit Sony GPU opened up new dimensions of creativity – literally – which brought new popularity to FPS shooters like Quake II that made stalking, shooting, and dodging more thrilling in 3D.

Each of these technical shifts produced a revolution in creativity that reframed how players game. Generative AI, however, could potentially drive the greatest advancements in video games since transistors made the idea of playing tic-tac-toe on a computer possible.

We’re at the hyper-fertile start of a new age with AI. Each week brings a staggering number of fresh ideas, demos, and services that simultaneously spark our imaginations and raise new concerns. And while those concerns are very real and will have to be recognized and addressed ethically throughout the ecosystem, the possibilities for generative AI in video game development are impossible to deny.

AI has played a role in game development far longer than ChatGPT has been on the scene. It’s already widely used to accelerate elements of graphics production, train simulations through machine learning, and automate testing and repetitive tasks. The future is limitless, but we can already start to see some applications take shape.

Every single aspect of game creation has the potential to be impacted, accelerated, and possibly improved by AI.

• Asset generation: Generative AI can be used to create game content like characters, terrain, vegetation, lighting, and even audio automatically, rather than being programmed by a developer. These capabilities can not only save development time and resources, but also enable previously unthinkable permutations during game play. Environments, for example, might feel less static or hard-coded to offer more immersive game experiences. The tech for this is already close – learn more about Unity Muse and apply for the closed beta to get started now.
• Non-player characters (NPCs): Interaction with NPCs during game play has been extremely limited by prescriptive programming behaviors. Generative AI may soon be able to power more lifelike and realistic NPCs that behave with greater adaptability and intelligence, for example by reacting realistically (yet unpredictably) to a player’s actions.
• Greater safety: A consistent issue with massively multiplayer online games (MMOs) has been unwelcomed toxicity and overall safety, especially for younger players. Coupled with natural language processing (NLP), generative AI can both more effectively manage the detection of harmful language and interactions to power safe yet entertaining interactions. Discover how machine learning is already helping devs to monitor in-game comms and build safer player communities with Safe Voice from Unity Gaming Services.
• Game design: Imagine if generative AI can be used to create game mechanics, levels, and character play that responds to player behavior and dynamically adjusts the game’s difficulty level accordingly. In short, it could be used to produce bespoke gaming experiences tailored to specific players. Non-deterministic game design can also open the door to entirely new genres of games – imagine, for example, compelling, open-ended detective games that change and adapt with every move.

But this potential doesn’t end with creation – every aspect of the interactive runtime can be impacted and solved with a mix of compiled code and ML-trained AI solutions.

• Runtime inference: With Unity Sentis, designers can build game loops that rely on inference – the process of feeding data through a machine learning model – on devices from mobile to console to web and PC, without cloud compute costs or latency issues. This will be used to run NPC characters like Orb or restylize a game without requiring all-new artwork (for a night scene, for example, very much as Hollywood does it), or it could replace a physics engine by something 1,000 times more efficient. This could be fluid dynamics or particle interactions – the use cases of a performant runtime-based inference engine are endless.

Generative AI will change our world in profound ways, known and unknown, but it will certainly empower teams to create entirely new game genres that will surprise and entertain us as never before. I am looking forward to the next few years to see what new genres are going to appear.

At Unity, we believe the world is a better place with more creators in it, and AI will be a powerful tool to help creators of all kinds unlock new dimensions of productivity and creativity. We can’t wait to see where you take us next.

Stay tuned to the blog for more about Unity and AI, and, if you haven’t already, sign up for the AI Beta Program to be the first to hear about new tools and services.