This essay is part of a series. World Creativity and Innovation Day 2026: Spark and Shield
Policy discussions around artificial intelligence (AI) have largely ignored the video game industry. This omission is unacceptable, given that the global gaming market will generate more than US$180 billion in annual revenue in 2025, more than the global film and music industries combined. It is also growing rapidly in regions that are high on the development agenda, such as Asia-Pacific, the Middle East, and Latin America. Game studios were among the earliest commercial adopters of pathfinding algorithms, goal-oriented action planning (GOAP), behavior trees, and procedural generation, often deploying AI technologies at scale years before they gained traction in other fields. AI researchers have long used gaming environments as testbeds for reinforcement learning and agent design. Rather than being adjacent, the two fields have been structurally intertwined for decades.
The relationship between gaming and AI is structural, built into the hardware that both industries rely on.
The relationship between gaming and AI is structural, built into the hardware that both industries rely on. Founded in 1993, NVIDIA developed graphics processing units (GPUs) to meet the rendering demands of video games. For over a decade, the GPU market has been driven almost entirely by gamers seeking faster frame rates and richer, smoother visual environments. However, the parallel processing architecture of GPUs turns out to be very well-suited for the operations underlying machine learning. By the early 2010s, researchers began repurposing gaming GPUs for neural network training, and NVIDIA pivoted to serving both markets. The dynamic between gaming and AI is one of co-evolution rather than parallel industries that occasionally intersect. The recent introduction of generative AI technologies in game development has accelerated another evolutionary dynamic that may impact labor markets, the creative economy, and the global distribution of technological capabilities.
long alliance
AI and video games have been intertwined since the industry’s inception, but the nature of that intertwining has changed with each generation of technology. In the 1970s and 1980s, games like Pong and Pac-Man relied on simple rule-based logic to control enemy behavior. Although these systems were rudimentary, they helped establish the principle that software could simulate decision-making in real time. This principle would expand dramatically in the following decades. By the 1990s, developers employed finite state machines (FSM) to give enemies and allies in games more realistic movements and reactions. title like golden eye 007 (1997) demonstrated how enhancing the intelligence of NPCs (non-player characters) through a relatively simple FSM can change the player’s experience of the game world, imbuing the game world with a density and unpredictability that cannot be replicated in linear, scripted encounters.
The 2000s brought us Behavior Trees and GOAP. games like fear (2005) famously used these architectures to enable layered NPC behavior. fear Of particular note is the enemy AI, which evokes tactical dynamism rather than merely reactive movements. Procedural content generation has matured in parallel with titles such as: minecraft (2011) and no man’s sky (2016), instead of duplicating or handcrafting all assets, they use algorithms to generate worlds, generating nearly infinite variations from a finite set of rules. Running through these developments was the fact that the gaming industry served as a laboratory for overlooked but important AI technologies, stress-testing them at a commercial scale under conditions that academic research environments cannot provide.
Generative AI as a new toolkit
Traditional game AI could perform actions, but could not generate content. Generative AI introduces qualitatively different features that cannot be inferred as a single phenomenon. Asset creation uses tools that leverage diffusion models and large-scale language models to generate concept art, texture variations, and drafts of 3D models much faster than humans. These are tasks that previously required hours or days of expert work. For narrative design, an extensive language model enables dynamic NPC interactions that respond to player input depending on the situation, rather than cycling through a fixed dialogue tree. The result is the potential for emergent storytelling that adapts to individual playthroughs. This has been discussed theoretically in game design for many years, but has only recently begun to become technically feasible.
As human capital requirements for game development shrink thanks to AI, questions around market access and industrial geography far beyond the gaming sector are becoming more important.
Worldbuilding would also benefit from similar advances. Generative systems can procedurally generate terrain, architectural, and environmental details with consistency and aesthetic qualities that established algorithmic approaches have struggled to achieve. Rapid prototyping may be the most transformative application in the near term. Small teams or even individual developers can now create functional game prototypes in a much shorter time frame.
The important distinction for policy is not between AI-assisted and non-AI development. That boundary is becoming obsolete. More importantly, as human capital requirements for game development are reduced by AI, market access and industrial geography questions that extend far beyond the gaming sector become more important. When the barriers to entry into game development are lowered, the question becomes who is in a position to step through those doors, and that’s a question with geo-economic dimensions.
Labor and industrial dynamics
The productivity gains that generative AI brings come with unpleasant but inevitable consequences. Some of the work currently done by human experts will be automated or re-engineered. Concept artists, texture designers, quality assurance testers, and junior writers are the roles most likely to be replaced. The generation tool can now approximate the output for these locations even if they are not already matched. The gaming industry will lose an estimated 14,600 jobs in 2024, exceeding the 10,500 recorded in 2023. These layoffs are not solely due to the adoption of AI, rather than other factors such as post-pandemic over-corrections or higher production costs, but it is hard to ignore the coincidence with the acceleration of AI adoption.
While the skills gap is real, AI workflows are still new to most sectors, meaning latecomers aren’t as far behind as they might assume.
Along with these losses, new categories of jobs are being created. Rapid engineering, integrating AI pipelines, curating training data, and monitoring AI-generated content all require human judgment and technical fluency that the tools themselves cannot provide.
Studios that once hired artists who could draw digitally now want artists who can direct AI models and refine their output to avoid generic, low-quality output, or AI slop. This is a hybrid competency, combining both aesthetic judgment and technical orchestration, to which training and education systems must be aligned. This transition truly opens the door for emerging markets looking to build a competitive game development workforce. While the skills gap is real, AI workflows are still new to most sectors, meaning latecomers aren’t as far behind as they might assume.
conclusion
The relationship between AI and the video game industry is entering a different phase, not in degree but in kind. For decades, AI has been a component of in-game mechanics. It is now becoming a component of development studios. This change has implications for labor markets, creative industries, and the global distribution of developing capacities that policymakers have been slow to take into account. Part of the reason for this is that gaming continues to be treated as a consumer entertainment product rather than as an important economic sector.
Countries and businesses that act now to create enabling conditions, rather than waiting for it to happen, will be in a position to capture the full value of this transformation.
Policy responses need to move beyond the dichotomy between unconditional acceptance and instinctive rejection. Players, developers, and policymakers bring legitimate but competing interests to the table, and lasting results will depend on a framework that takes all three seriously, rather than privileging one at the expense of others. The opportunity is real, especially for emerging markets, but it is conditional. It depends on deliberate investment in infrastructure, skills and legal institutions, not just the availability of cheap tools. Countries and businesses that act now to create enabling conditions, rather than waiting for it to happen, will be in a position to capture the full value of this transformation. Those who treat games as peripheral to AI challenges will soon realize that they have misunderstood both.
Siddharth Yadav He is a Fellow in Emerging Technologies at ORF Middle East.
The above views belong to the author. ORF research and analysis is now available on Telegram. Click here to access our carefully selected content (blogs, long-form articles, interviews).
