GenAI at devcom/Gamescom:
Reshaping Game Development Through Automation, Process Enhancements & Personalization
As the use of AI becomes more widespread, it’s no surprise that its application within the gaming world dominated the agenda at devcom, the official game developer conference of Gamescom, held in August 2024 in Cologne, Germany.
In light of this, Jeff Skelton, Head of Technology Partnerships, Electronic Arts (EA) sat down with Vitalii Vashchuk, Head of Gaming, EPAM to discuss how Generative AI (GenAI) is reshaping game development. The two gaming leaders discussed how GenAI can streamline workflows, increase efficiency and reduce production timelines, empowering studios of all sizes to develop high-quality games faster and effectively.
Watch the session below or read our blog for key insights shared during the session.
4 Key Areas of GenAI Application
To kick off the session, Vashchuk asked about exciting applications of GenA in gaming.
Skelton explained that when it comes to applying AI, EA likes to take a human-first approach. The priority is enabling and enhancing creativity, providing an environment where developers can get into their creative flow and stay there.
“It's all about getting barriers out of the way and letting them give the best that they can to our players,” explained Skelton.
AI and GenAI can empower talent to focus on creating incredible interactive experiences. Skelton shared that EA provides tools and techniques to remove unnecessary or mundane tasks and enable creativity.
“[AI] lets us do more and do better — and let people spend more time exhibiting their creative genius and doing what they're good at, which is creating art or creating code, not doing emails or spending time in meetings.”
In terms of how AI is being used, Skelton said it comes down to four key areas:
- Tooling
When it comes to tooling, AI is being used to support more efficient content creation, curation, coding and iteration times. The tools enable engineers to iterate faster, artists to iterate on concepts and ideas faster and designers to get ideas down into something that can be adapted and developed faster. - Runtime
Small AI models are already being deployed inside games for things like physics and animation and rendering. EA has been doing this for a number of years. However, coming down the road will be the big ones — the LLMs, Speech to Text (STT), Text to Speech (TTS), Non Player Characters (NPCs) with memory and personality. These are the things we don’t have the power to do at the moment, but the whole industry is looking at what is possible.
One of the most exciting opportunities is natural language interface — not just for developers but for players, where they can talk to the game and the game will semantically understand the concepts they want. Instead of players using multiple sliders to specify the character they want, they can just describe it verbally. AI will make it and then they can just speak to iterate on it. Developers will be able use the same capability to quickly prototype things.
AI is also going to impact User Generated Content (UGC). Players will be able to create content to put into games, giving them the ability to tell the game what they want
Accessibility is another area where AI will make a huge difference. Game assistance can be developed to enable spoken language assistance for the sight impaired. Lookup tables for color blindness can be developed at speed. These things will be done on the fly using AI, whereas now they have to be hard coded so they are not always made available.
“Out of all of this, we're going to see huge, realistic worlds with hyper personalization in them, which has been hard in the past.” - Testing
AI can help automate testing, interpret the results and suggest fixes. The use of intelligent agents and tools enables human testers to speed up their work.
Together, humans and machines generate huge volumes of test results. AI helps analyze those results, identify bugs, suggest fixes and also make the change in code, run that change through smoke tests and share the results. Developers won’t have to stop what they’re doing, go fix a bug and come back — AI will present a potential fix that has been tested so all that needs to be done is to push the change live. That’s massive in terms of productivity. - Productivity
Regardless of industry, AI has made a huge impact on employee productivity — whether it’s email or scheduling assistance or semantic search of source code. At EA, the team is using Retrieval Augmented Generation (RAG) tools to give employees the ability to tell it who they want to meet with and for how long, and it will find the best available time. But coming soon, employees will be able to simply hit a button, and it’ll make the meeting.
Personal agents are another area offering huge productivity gains. For example, if a crash occurs, a series of agents will identify where it happened and the cause. But they will also be able to identify a fix, check the relevant file out, make the fix, shelve it, initiate pre flights and smokes, ensure they work and then email the relevant person to detail what has happened and what’s been done. Then all that person needs to do is verify the change. We’ll be able to do the same thing with data pipelines.
Skelton and Vashchuk explained that both EA and EPAM have set up similar programs that include extensive internal infrastructures, so it can quickly and easily explore different models.
Enhancing Game Design
EA has been using AI for more than 20 years as part of its game design and construction, with over 160 AI projects running currently. Concepts can be developed quickly and explored at the speed of thought using native language, speeding up the whole process.
In fact, EA used AI to create 150 stadiums for EA Sports College Football, whittling the stadium creation time down from 6 months to 6 weeks. Additionally, it has improved head creation. In the past games have had a mix of fidelities of heads in a game. Using AI, a 2D image of a player can be used to generate a medium fidelity 3D head with likeness. This meant 11,000 heads were able to be included in the game and were of a much higher fidelity than what would’ve been possible previously.
“It’s about enabling a massive amount of content to get into the games, stuff that couldn’t be done before.”
Data & Integration Challenges
Though AI offers many benefits, it doesn’t come without its challenges. Connectivity is key, as is managing cloud costs, which can quickly escalate. Latency remains a constant challenge across the industry. Collectively, the industry needs to look at device inference. Then there is the min-max problem. If AI is to do gameplay mechanics, it has to work on every device that can play that game. And every device is multiplayer, which can span eight or nine years’ worth of hardware. Getting AI onto games on all of these devices is going to take time. Models will need to be small or limited to non-gameplay features (like voice, rendering or particle effects).
AI Will Continue to Change the Game
What is clear following this discussion at devcom is that AI is being used extensively across the gaming industry from content creation, coding and concept development to testing and design. Its use has already increased efficiency and productivity as well as improved personalization. For EA, the company has been able to use it to help employees spend their time doing what they do best: delivering improved player experiences. AI will continue to expand and deliver significant improvements across the the industry — ultimately, it will continue to revolutionize gaming for years to come.
