AI and mobile games: Roblox chief scientist Morgan McGuire

Imagine coding a game by voice. As in, telling a smart AI system what you want your game to be, how you want it to look, and how gameplay should happen. And then the smart AI system just goes off and writes code, compiles the software, and provides a nice publishable package. That’s just part of what we’re going to see in the future of AI and mobile games.

We’ve seen early bits of that already. 

But it’s going to get mind-blowingly incredible.

I recently chatted with Roblox chief scientist Morgan McGuire about the future of AI and mobile games. Hit play, and keep scrolling …

“I think we’ll get to the point where in the future you could, if you wanted, make a game entirely with voice. So by talking to the system, working with it over the course of an hour,” Morgan McGuire, chief scientist at Roblox, recently told me. 

“It’s just up-leveling what you’re doing. Instead of typing every single thing using your mouse, we can have AI take your intent and produce it for you.”

And it’s likely not far away.

AI and mobile games: the potential

AI is increasingly being used in game development for almost everything. And the potential for what game publishers can do with AI is off the charts. Here’s just a few examples:

• Content generation
  Maps, phrases, ideas
• Game management and liveops
  Segmentation, offers, challenge levels, player matching for battle games
• Smart NPCs
  Way better team play even in solo mode, as the former Unity CEO also highlighted.
• Endless levels or world-sized environments
  Larger environments and longer games with less work
• Art and design generation
  Character, clothing, accessories, weapons …
• In-game communications
  If you’re going to allow players to chat in a global game, real-time translation would be handy. And if you need to moderate that — with humans or AI — you need to understand multiple languages.
• Game development
  Building game components and eventually full games with AI
• Game analytics and monetization
  Optimize player progress while monitoring in-app purchases and offering the right upgrades
• Cheat detection
  Finding players or bots that are not playing according to the rules

Of course, not everyone is using AI for everything. The tools are mostly new and still growing into proven everyday applications. But almost every marketer is using AI already, and most developers are too.

Some, however, have almost no choice but to rely heavily on AI.

AI and mobile games: what Roblox is doing

Roblox fits into that category for a lot of reasons. 

First, Roblox’s aim is to connect a billion people in a safe online environment. It’s already massive, with 380 million monthly average users, which is roughly 2X Steam, 3X PlayStation, 5X Minecraft, and 2.25X Fortnite. In fact, Roblox is so big that VC and game-maker Matthew Ball says it’s likely that Roblox has more monthly users than the entire AAA gaming ecosystem combined.

Keeping something that size that safe and moderated — keeping out the predators so the kids can enjoy themselves and not get targeted — is a mammoth task. Especially with more than 50 languages in use on Roblox, in an environment that allows real-time chat and voice communications between players and sees more than 50,000 text-based messages a second

Essentially what this is an AI-sized task. Which is precisely what Roblox has done.

“We have the world’s first voice moderation system that will monitor 100% of voice chat on Roblox,” McGuire told me. “This is deployed.”

Real-time translation. Real-time moderating. And done at a level of quality — not just quantity — that surpasses human capabilities, McGuire says.

But Roblox doesn’t only have a content moderation problem. Like any other successful game maker, Roblox has a content problem. 

A big one.

Every month, people spend a staggering 6 billion hours on Roblox. Every day, people spend time in more than 5 million active experiences (think games, environments, or social spaces). All of that needs building, refreshing, maintaining, updating. And without constant development of new and exciting experiences, players might go elsewhere.

After all, Fortnite is a tap away. So is Minecraft.

“As the user base grows, we need the content to grow. It has to grow roughly proportionally,” says McGuire. “Going back for roughly the last 15 years, the games industry and the film industry had exactly this problem: they couldn’t produce content fast enough. They were victims of their own success.”

The solution: generative AI.

100% of Roblox content is community built. And while there’s been tools to help build it, it’s still 3D modeling and programming. Even with the friendliest and easiest language on the planet, not everyone is going to get into writing code and developing games.

But with voice input to an AI game generation engine … suddenly anyone can.

“AI is what enables us to lower the barriers to creation,” McGuire says.”So that the people with great ideas, the strength of their ideas will shine through and they don’t have to spend one year, three years, five years learning how to program, learning how to do 3D modeling, learning how to do rigging animation.”

Building a game from scratch by voice isn’t there yet. 

But Roblox has already unveiled generative AI tools for image generation — think game accessories, avatars, clothing, etc. — and for what Roblox calls 4D generative AI. Those are generative tools that don’t just make discrete images but game-ready components with an awareness of 3 dimensional spaces, physics, and how to interact with avatars.

Now AI isn’t just making static objects, but it can also bring them to life within interactive, moving 3D spaces.

The ultimate goal is to match the content consumption stat — 100% of Roblox content is community built — with a content creation stat: 100% of Roblox users create new content.

AI and games: what’s missing

There’s a big missing puzzle piece in growing AI in mobile games, however. AI is already saving Roblox millions of dollars a year in human moderators it doesn’t have to hire, but it’s not enough.

AI isn’t efficient enough. 

It costs too much in terms of compute and energy.

So if Morgan could wave a magic wand, it would be to solve the efficiency challenge in AI processing. His ultimate goal: to achieve machine efficiency about 10 times better than what we see today, paving the way for more sustainable AI applications. 

“Every time we make something more efficient, we make our business better,” he says. “We monetize better, our profit margin gets better … but we’re also consuming fewer resources.”

It’s a trillion-dollar opportunity, McGuire says, for someone who could achieve it.

And the result will be more new … everything.

“A million people are making new ideas on Roblox every day,” he says. “And if one of them is a winner, then it’s great for all of us. We all get to experience that new thing.”

Much more in the full podcast

As usual, there’s much more in the full podcast with Roblox chief scientist Morgan McGuire. Check it out, subscribe to Growth Masterminds, and get all the insights.

Find all the links you need to subscribe right here …