Train AI with video games: How can virtual environments facilitate AI development?

• Author:
• Author name

  Quantumrun Foresight
• February 27, 2022

Insight summary

Harnessing the power of video games, artificial intelligence (AI) is being trained to enhance its collaborative and self-learning abilities, simulating real-world tasks and challenges. This approach not only aids in personalized learning and education but also offers industries, from healthcare to disaster response, tools for quicker and more efficient solutions. As a ripple effect, this trend could reshape the entertainment industry, introduce new job roles, and influence the way businesses and governments operate.

AI training through video games context

For years, computer algorithms have dominated humans in 1v1 games like chess, but getting AI to perform well with teammates has been quite challenging. Thankfully, using video games to train AI in collaborative scenarios may help solve this problem. Video games are designed (among other goals) to challenge the brain.

Players constantly learn new tactics, navigate different settings, work together with teammates, and overcome obstacles as the difficulty increases with each level. Similarly, training AI algorithms in these virtual environments can enhance their self-learning and collaborative abilities, reducing development costs and saving valuable time for businesses. In addition, video games can easily simulate real-world tasks and train computer algorithms for similar applications. 

For example, Google's DeepMind conducted a study using 30 neural network algorithms which teamed up in a capture-the-flag platform known as the Quake III Arena. Through deep learning, the bots exhibited high-level behavior and strategies that real players use in-game. Hence, strategy-based, open-world platforms like Minecraft, Starcraft, and Grand Theft Auto can help AI develop more valuable skills such as navigation, faster reaction times, planning, management, and creative foresight. In addition, game statistics and achievements can provide crucial insights into the algorithm's development and help identify where its specialization lies. 

Disruptive impact 

Training AI using video games can have a profound impact on individual skill development and education. By simulating real-world scenarios in a controlled environment, video games can be used to train AI models that can then assist in personalized learning. For instance, an AI trained in a game-like setting could be used in educational software to adapt to a student's learning pace and style, offering tailored lessons and challenges, leading to more effective learning experiences and better retention of information for students.

Beyond the automobile and gaming industries, companies in sectors like healthcare could benefit from this trend. Imagine training an AI in a medical simulation game, where it learns to diagnose diseases based on symptoms. This AI could then assist doctors in real-life diagnoses, making the process faster and potentially more accurate. Similarly, companies involved in disaster response could use AI trained in crisis simulation games to predict and manage real-world emergencies, ensuring quicker and more efficient responses.

For governments, using video games as training grounds can help AI manage complex logistical tasks, such as urban planning or resource allocation during crises. For example, an AI trained in a city-building game might provide insights into optimal infrastructure development or public transport routes. Additionally, defense departments could use war simulation games to train AI models, which could then assist in strategic planning without putting human lives at risk. 

Implications of training AI with video games

Wider implications of AI training through video games may include:

• Increased sales for custom AI systems, further expanding the number of creative and technical jobs in the machine learning industry.  


• Reduced general costs involved with AI development and accelerating AI adoption in industry.


• Machines engaging in more human-like conversations through natural language processing (NLP).


• Service robots that become capable of performing repetitive and navigation-based tasks, as well as operating safely in environments populated by humans. 


• New job roles, such as AI trainers and virtual environment designers, while reducing demand for traditional game developers.


• The entertainment industry shifting its focus from traditional storytelling to creating dynamic, AI-driven narratives, leading to a more personalized entertainment experience for consumers.


• New regulations to ensure the ethical use of AI in video games, leading to a more controlled and safer gaming environment.


• Reduced need for human players in multiplayer settings, leading to a more solitary gaming experience and impacting social interactions among gamers.


• Companies adopting a subscription-based model for games with AI-driven content updates, leading to consistent revenue streams but increasing costs for consumers.


• The environmental impact of running advanced AI models increasing the demand for energy-efficient gaming hardware, leading to a shift in manufacturing priorities for tech companies.

Questions to consider

• Do you think that more companies need to implement AI in their business infrastructure? Why?


• What other forms of AI training would benefit from video game training environments?