Google Cloud expects live services to evolve into AI-based living games | Jack Buser

Live services have helped games continue to generate revenue for a long time. But Jack Buser, director of game industry solutions at Google Cloud, believes that AI advances will lead us to something better: Living games.

Living games expand the live service model to include the game itself as an active participant, Buser said in an interview with GamesBeat just ahead of the Game Developers Conference next week. It’s as big a change as when the CD-ROM arrived decades ago and enabled games with rich graphics like Myst and The Seventh Guest, Buser said.

“We’re calling this ‘living games’ and we may have hinted at this to you before. But it’s really taken shape over the past six months. And it’s something that we’re incredibly excited about,” Buser said.

You can imagine virtual worlds that dynamically adapt based on player choices, generate unique quests and challenges, and continuously evolve their narratives. This fosters a deeper level of player immersion and engagement, leading to a more enriching experience for everyone involved. In other words, the games truly adapt to the things that you do in the game.

“As you know, over the past few years, we have built out an ecosystem for live service games, which contemplated game servers, databases and analytics. And our message to the industry was to use battle-hardened, tested solutions for your live service backend so you can focus on building a great game,” Buser said.

He added, “We would make the same tools that we use for our own live services at Google, and [we would make them] available to the games industry so that they can focus on building great games,” he said. “It’s ready for massive scale and massive success.”

Enter generative AI.

“But over the past year, and specifically over the last six months, we’ve seen a tremendous amount of momentum in the industry such that we’ve actually coined a new term, ‘living games,’ to try to describe what we see happening in the industry and what we believe games are going to become over the next few years,” Buser said. “If you think about how a traditional live service game, the idea is that there’s this feedback loop between the player and the development team. The player either tells development team what they want, or the development team has an understanding of a particular cohort of users and then creates more content for the game to keep them entertained.”

He added, “What we’re seeing moving forward is actually that loop will have another endpoint, which is the game itself. And so now in the future, you’re going to see this idea of a living game that can then respond to the needs of the players using generative AI and provide gameplay experiences we simply haven’t seen before.”

In other words, the AI will power the game, and it will provide things that players want, without the need for the game developers to specifically program things to make it happen. This is not cutting game developers out of the picture. Rather, it is freeing the game up to respond to the players and freeing the developers to focus on other aspects of making the game great. You can think of it as automated live services that will keep players entertained.

You will create feedback loops from the game, which can be controlled by AI, to the human players without the need for the game developers to intercede. This kind of thing used to be impossible, but generative AI and smart non-player characters will enable it, Buser said.

Keeping things under control

To ensure living games do not generate inappropriate or misleading content, Google Cloud offers customers a grounding feature, which helps reduce hallucinations that sometimes pollute the results coming out of an AI model. Grounding a foundation model helps to better equip it to understand language and other abstract concepts in the context of the real world, Buser said.

Buser believes that the key elements of living games include live services, AI LLMs, and AI tools such as voice generation and automated production. That certainly sounds like a technology-heavy agenda for the future of games, but then that’s what you would expect from Google Cloud.

LLMs are coming to games

Global game developer and publisher of popular titles including Guild Wars 2 and Blade & Soul, NCSoft, is using Google Cloud’s AI infrastructure to power its in-house large language model (LLM) set, Varco LLM, Buser said.

NCSoft’s Varco LLM is a model specialized for generating high-quality content required for game development. Using Varco LLM, NCSoft has developed a suite of AI tools including text creation, managing “digital” human characters, enabling player-to-virtual-character conversation, and dynamic storyline generation based on player actions. The result is a more engaging and dynamic gaming experience for players, Buser said.

Building the future of play through a collaborative ecosystem

The future of games is also collaborative, Buser said. By fostering a thriving ecosystem of game developers, technology providers, and players, Google Play and its partners are building truly living worlds that push the boundaries of storytelling and engagement. 

Google Cloud goes beyond just offering its own technology and infrastructure. The company acts as a bridge, connecting developers to the most powerful and innovative game development tools from industry leaders. 

Google Cloud is also partnering with GlobalLogic, owned by Hitachi, to enhance game production, gameplay, and distribution.

Extending live services

The elements of live services that can be used for living games are the ability to track player data and getting insights into player behavior and what they like and how they engage with the game. In living games, developers will be able to evolve content and personalize narratives to keep content fresh for gamers.

When it comes to personalization, games already use analytics to understand players. Now the AI could use that knowledge to dynamically generate content in the game based on that understanding. Currently, in the live service model, teams have very sophisticated analytics stacks, often on BigQuery from Google, where they can understand their players and then create content for those players, and then deploy those that content for those cohorts and players.

“But imagine if the game could use that same understanding to dynamically create content itself. For that audience. That’s incredibly exciting to us,” said Buser.

That means that the AI could offload the task of merchandising items toward players, and then the developers can go back to focusing on the gameplay.

For example, the recommendations capability of Vertex AI Search can personalize in-game recommendations for items, quests, or content based on a player’s past behavior and preferences. Devs can also understand player behavior through robust data analysis using BigQuery and Looker.

These tools allow developers to tailor content and features for maximum engagement. Imagine a game dynamically adjusting difficulty based on a player’s performance tracked in BigQuery and visualized in Looker, Buser said.

By providing a powerful, adaptable, and secure platform for live service games, developers can focus on what they do best: crafting exceptional games and building thriving player communities, Buser said.

Now that generative AI has arrived Buser believes the tech empowers creators – designers, developers, artists, and marketers – to reimagine their workflows and craft experiences that were previously unimaginable. The live service model, while successful, operates on a developer-player feedback loop. The future points towards a more collaborative approach – living games, he said.

Will living games be more popular than live services games?

This is probably a good point to note that a lot of gamers have expressed frustration with live services, as Warner Bros. Games learned when an exec said it would consider more live services games because of weak sales of the single-player triple-A game Suicide Squad. Players are concerned that live services are designed to continuously extract money from them on games that they’ve paid for.

Google Cloud argues that player engagement is at an all-time high, and a significant portion of that comes from live service games.

A recent survey paints a clear picture: a staggering 95% of studios are developing some form of a live service experience. In fact, Electronic Arts (EA) reported that live services comprised more than 73% of its business in 2023, even while EA continues to release traditional retail games.

Google Cloud said this is a testament to the power of this evolving model and signifies a crucial shift – developers understand the need for games that evolve alongside their passionate player bases. But success is not a foregone conclusion for those who think live services are coming to the rescue.

Since “live services” has become a loaded term, Google Cloud wants to start talking about a new term with “living games,” Buser said.

“A live service game can mean something very specific to game players and game designers. In fact, when we would talk about that term to a particular games company, often they would take a very narrow view of the term based on the types of games that they build, and rightfully so,” Buser said. “At Google, we always took a very broad view of what a live game could be. Like, even if you look at the biggest single-player games on the planet, where you would say that absolutely not a live service game. If you look closely enough, you will see live service modalities in the game.”

As a result, Buser argues you would be really hard pressed to find a game that is a top selling game, but has no live service aspect whatsoever.

“There is a lot of mixed opinion in the industry about the term, even with the players. But if you look at the actual facts, these are modalities that every game studio on the planet needs to understand how to how to do and players simply expect it,” Buser said.

Will AI win players over?

Google Cloud has also teamed up with Replica Studios, an AI voice technology company.

Players are also not so enamored with AI. OpenAI’s AI solutions got booed at SXSW last week. And Replica Studios faced a backlash from actors when it struck a deal with SAG-AFTRA, the union which represents actors, for the licensing of actors’ voices for use in video games. But actors objected to using a digital replica of their voices for game development and other interactive media projects.

Replica said that its deal with SAG-AFTRA showed it was taking an ethical approach to AI in game development. Replica’s Voice Lab voices are sourced from Replica’s library of contracted actors authentically representing a range of characters available to the platform’s user. 

Buser said that Replica Studio’s Voice Lab technology breathes new life into non-player characters (NPCs) and stories with advanced, cost-effective generative AI. This will help power living games, he said. Replica’s Voice Lab takes cues from a prompt to generate a unique voice, and when combined with Google’s Gemini Pro model, can utilize a game’s assets, environment, and lore to create a cast of thousands of diverse characters with unique voices, personalities, motivations, and behaviors. 

“This is where the living game concept really takes place. It’s when the generative AI algorithm isn’t necessarily being accessed by a professional developer, but it’s being accessed by the game player themselves directly. So you can create feedback loops now between the player in the game without a human on the development team necessarily even involved in that loop. And this creates experiences that simply would be impossible before,” Buser said. “Some of the first ones we’re going to see out of the gate are things like smart NPCs. We are working very closely with Replica. The idea that you can create living breathing characters in a game that aren’t necessarily pre-scripted, and are fully voiced and ready to go, and can create a much richer, more interactive world for players is an incredible first step towards this concept of a living game.”

Partners like Replica Studios, Globallogic and Google Cloud are ushering in a new era of living games that will define the next generation of interactive entertainment, Buser said.

“A challenge I hear most from game developers is scalability, especially in massive open-world games,” said Shreyas Nivas, CEO of Replica Studios, in a statement. “From believably populating bustling cities and markets in today’s most popular games, it’s meticulous work creating each character–even basic NPCs–sometimes in the thousands. We knew there was a better way, and we have the best partners in GlobalLogic and Google Cloud to show what’s possible at such a large scale.” 

Building on assets created by Synty Studios to enable game developers, GlobalLogic created a vast environment and story demonstrating the power of Replica’s Voice Lab, enabling dynamic player interactions and showcasing the strength of Google Cloud’s support for triple-A games using this technology. 

Buser said Google Cloud’s partner ecosystem provides access to the latest engines, online back-end solutions, Web3 platforms, and gen AI tools, empowering game developers to focus on what matters most: crafting experiences that resonate with players on a deeper level, fostering lasting connections and a shared sense of community within living games.

Buser said Google Cloud wants to empower the next generation of game experiences. By embracing transformative technologies like generative AI and fostering a collaborative ecosystem, developers can unleash their creativity and bring to life games that are more immersive, dynamic, and deeply engaging than ever before, he said.

Auto-generated games

Google Cloud has also teamed up with Hiber, which has created a demo for using AI to generate games based on text prompts from users. You can use Hiber to create a world where you can try it out firsthand.

Buser described the Hiber platform as “incredible.”

He said, “It’s the first time the world has ever seen anything like this. There is still a long road ahead. But the fact that you can just type in, ‘I want to be on a desert island. With these things on the island,’ and it just makes it. That’s unbelievable. It doesn’t have to be explicit like that. It can also be implicit. It can also be the game just knows what you want. It just understands you as a player because it understands your cohort.”

The primary distinction between ordinary live service games and living games is really around the quality and sophistication of the AI, Buser said.

The future

Whether you’re as found of these technologies as Buser is, one thing is clear when it comes to the combination of live services and AI: we’re just at the beginning.

“Where does all this go? We believe that ultimately, this goes to a place where game designers are going to build games that are foundationally, predicated on generative AI that they simply couldn’t have been built [before] had there not been generative AI,” Buser said.

He added, “Just like those other technologies, in the early days, you’re going to see existing game types start to integrate this new technology to dabble with some new functionality. But ultimately, we’re going to see games designed fundamentally with this technology in mind, just like we saw with 3D.”