A group of former Google DeepMind researchers has developed an innovative AI behavior engine that promises to transform traditional video games by making non-playable characters (NPCs) more dynamic and interactive. Artificial Agency, based in Canada, has emerged from stealth mode with $16 million in funding, betting that its technology will set it apart in the growing AI market for video games.
NPCs (Non-Playable Characters)
Traditionally, NPCs in video games have followed predictable patterns based on decision trees and pre-written scripts, limiting the variety of responses and behaviors. This often results in repetitive dialogues and unrealistic experiences for players. When these prefab decision trees are exhausted, no matter how much players interact, they receive responses like "Mmm..." or "Uh-huh...". Artificial Agency, based in Edmonton, Alberta, is challenging this approach with its behavior engine, which turns developers into directors. Instead of following rigid scripts, NPCs are given motivations, rules, and objectives that dictate their interactions with players, creating more natural and diverse responses.
Artificial Agency’s technology can be integrated into existing video games or serve as the foundation for new developments. The startup is entering a competitive market, facing companies like Inworld and Nvidia, which also develop AI-powered NPCs. However, it is confident that its innovative approach and collaborations with "several notable AAA game studios" will give it an edge.
The Future of AI in Video Games
The recent $12 million funding round, co-led by Radical Ventures and Toyota Ventures, adds to a previous $4 million round from Radical Ventures, bringing the total raised to $16 million. Other investors include Flying Fish, Kaya, BDC Deep Tech, and TIRTA Ventures. According to co-founder and CEO Brian Tanner, the interest in their technology is not about if it will be adopted, but when. Tanner highlights that integrating AI-generated behaviors in NPCs will be a sure bet in the video game industry in the coming years.
The adoption of AI in video game development still faces some doubts. According to a GDC and Game Developer report, nearly half of the 3,000 developers surveyed use generative AI in some aspect of their work, but only 21% expect it to have a positive impact on the industry, while 42% are concerned about the ethics of using generative AI. Daniel Mulet of Radical Ventures is confident in the founding team of Artificial Agency, given their experience at Google DeepMind, famous for developing AlphaGo, the first program to beat a world champion at Go. Mulet points out that many game studios are trying to develop this technology internally, but the opportunity to have a platform available to all could change the landscape.
From NPC to Cooperative Companion
In a demonstration of their technology in Minecraft, co-founder Alex Kearney showed how an NPC named Aaron could perform complex actions without being explicitly programmed. Aaron gathered supplies, responded to instructions, and adapted his behavior dynamically, creating a unique experience and saving developers time.
Although the five-minute demonstration cost a dollar in AI inference costs, Tanner expects these costs to decrease significantly in the future. The company uses open-source models like Meta’s Llama 3 and is confident that costs will drop with improved GPU efficiency and AI model optimization. Tanner anticipates that within a year, a similar demonstration could cost a penny or less. The question of who will pay these additional AI costs remains open. While Artificial Agency does not expect video games to become more expensive for end users, Mulet suggests there could be a monthly fee for players. "Will you, as a player, pay $2.99 a month or $12.99 a month? It's a bit early to tell," Mulet comments.
