Since the 1950s, artificial intelligence (AI) and video games have evolved alongside each other. Due to their clear objectives and structured rules, games have long served as ideal testing grounds for emerging AI models. This paper explores the historic and ongoing relationship between AI and games, first describing milestone achievements due to AI victories in Checkers, Chess, Jeopardy!, and Go. Beyond just testing its capabilities, games have also integrated AI into gameplay and development, which is showcased by non-player characters (NPCs) with sophisticated behaviors and AI-enhanced graphics. Modern developments also show artificial intelligence improving player experience through personalized adjustments and real-time assistance. Additionally, AI streamlines game development by assisting in content generation, debugging, and performance optimization. Looking ahead, this tool has the potential to make games more immersive, accessible, and complex. As the gaming industry continues to grow, AI will play a central role in shaping its future, pushing the boundaries of realism, interactivity, and inclusivity.
Introduction
With the recent boom in popularity and demand for software integrated with artificial intelligence (AI), it’s easy to believe that this tool is a recent technological development. However, AI and video games have been evolving side by side for more than 70 years [1], [2].
Games have long served as a testing ground for AI due to their rule-based systems with clearly defined win and loss conditions, which make it easy to tell whether or not an AI system is performing efficiently [1]. Winning a game also requires the ability to make complex decisions quickly, making it a perfect challenge for AI systems [1]. Now, AI and games still share a meaningful connection, given that artificial intelligence can be used to maximize player engagement, govern the behaviors of non-player characters (NPCs), and improve the overall game development process [3], [4]. This relationship is predicted to continue advancing, making the future of gaming more immersive, realistic, and accessible to a wider range of players [5], [6].
The Evolution of AI Through the Lens of Games
Figure 1. Arthur Samuel with Samuel Checkers. [1]
Games have long been used to measure the success of artificial intelligence [3]. Throughout the years, many milestones in the development of AI have been showcased through them[1], with Samuel Checkers being one of the most notable[1], [7], [8]. In 1949, Arthur Samuel joined IBM and began developing a checkers program –later known as Samuel Checkers– that could analyze moves ahead of time and make decisions while playing, essentially acting like a human player [1]. Samuel incorporated a process by which his checkers program could learn from games that it had already played [1]. He called this “machine learning,” a process that later became revolutionary and integral in the development and evolution of AI [1].
By 1997, IBM had produced a supercomputer named Deep Blue [7]. In May of 1997, world chess champion Garry Kasparov played a high-stakes chess match against Deep Blue and lost [7]. One of the greatest achievements in AI advancements since the 1950s, this match was recorded in history as the first time a computer had defeated a world chess champion [7].
However, AI’s achievements didn’t stop at board games. In 2011, IBM’s Watson DeepQA computer made history in a Jeopardy! contest when it beat two all-time champions, Ken Jennings and Brad Rutter [8]. Watson’s victory demonstrated major progress in AI’s ability to understand, interpret, and respond to human language—an area known as natural language processing [8].
These former chess and Jeopardy! champions wouldn’t be the only ones to see defeat against AI opponents [7], [8]. In 2016, world Go champion Lee Sedol lost against DeepMind’s AlphaGo software [7]. Chess is played on an 8×8 board, while Go, on the other hand, is played on a 19×19 board, where the number of possible moves per turn in chess ranges in the tens, in Go it ranges in the hundreds [9]. AlphaGo won by analyzing the best play positions at any given moment instead of considering all possible moves [7]. This victory for AlphaGo highlighted advancements far beyond what Deep Blue was able to achieve in its time through the use of deep learning, a type of machine learning based on neural networks[7].
The Integration of AI in Video Game Characters
Throughout these milestones, games were used for more than just testing the abilities of artificial intelligence. AI was also being integrated into mainstream gaming experiences, such as everyday video gaming [2].
Figure 2. Pong (1972). [10]
The first mainstream video game to make use of AI was Pong (Figure 2) [11]. This retro classic simulated table tennis, and players could either compete against each other or against an AI, with the latter being used to control the opponent’s paddle, creating a competitor for single players.
Figure 3. Pac-Man (1980). [12]
Many classic games like Space Invaders (1978) and Pac-Man (1980) also relied on AI for governing opponents’ behaviors [2]. This allowed for challenging and dynamic gameplay unlike anything that had been seen before [2]. Pac-Man, specifically, made use of AI to give each enemy ghost (Figure 3) a different personality, with differing behaviors and movements [11].
The ghosts of Pac-Man are an early example of AI-governed non-player characters, or NPCs. NPCs are an important part of many video games, but cannot be directly controlled or played. They can act as allies, opponents, or even just background characters. AI plays a large role in the development and behaviors of NPCs [2], [4].
Figure 4. Red Dead Redemption 2 NPC on Farm and in Bar. [13]
Much of the world of Red Dead Redemption 2 (2018), for example, is brought to life because of its rich NPCs [13]. To demonstrate this, one player decided to follow the game’s background NPCs around for a full in-game day [13]. He discovered that, despite them being side characters, these NPCs had complex routines [13]. One NPC (Figure 4) starts his morning off by leaving his house and heading to work on his ranch, first tending to the sheep [13]. A couple of in-game hours later, this same NPC finishes tending to the animals and begins moving around bags of feed [13]. At around sundown in-game, he stops working [13]. He then heads to the local bar and drinks with other NPCs before eventually returning home to sleep [13]. This example from Red Dead Redemption truly emphasizes the level of depth and realism achieved in games that have integrated AI into their NPCs [2], [4]. When AI is used to help govern NPCs’ actions and interactions, it allows them to behave more intelligently and in a responsive manner [4]. Their decisions simulate those of humans, making gameplay more adaptive and realistic [4].
The role of AI in games only continues to grow. AI now shapes gameplay experiences for players as well as game development processes, benefiting both gamers and developers [2].
Modern-Day Benefits of AI in Games
Figure 5. GTA V Before and After AI Photorealism Enhancement. [14]
Game developers today are able to make use of AI at many steps in the development process. For example, AI can be used to create and enhance game graphics to make them more visually stunning [2], [3]. This is done by combining AI with real-time ray tracing –a rendering technique that simulates how light behaves– to create realistic shadows, reflections, and lighting [3]. To further showcase AI’s potential to make games look photorealistic, Intel created an enhancement engine [14]. Their engine took scenes from Grand Theft Auto V, extracted information about the types of materials and lighting used in each scene, and scored how realistic they were [14]. It then enhanced these scenes to make them look more photorealistic (see Figure 5) [14]. Additionally, large language models (LLMs) can be used for efficiently generating content, such as procedural content generation being used to create diverse new game world environments [2], [4]. These advancements are helping to expand creative possibilities in the industry.
AI also benefits developers by speeding up the development process without the tradeoff of reduced quality [2]. This allows for games to be released quicker and in a more cost-effective manner [2]. Furthermore, AI can serve as a personal game tester and debugger for developers, detecting and solving issues at different stages of development [4]. This ability saves developers time and money and reduces the chance of releasing a game with major bugs or glitches that need to be patched later [4].
AI also plays a large part in cultivating more engaging and immersive experiences for players [2]. It has become capable of analyzing player data and applying this to create more personalized gameplay [2], [4]. One example of this is through player-experience modeling, which allows for a game’s difficulty to be adjusted in real-time according to the player’s skill level [4]. AI is also extremely valuable in creating smart opponents, as emphasized by its victories against Chess, Jeopardy!, and Go champions [3], [7], [8].
From delivering more personalized and responsive gameplay to reducing costs and improving efficiency for developers, the pros of AI’s role in games form an expansive list [2], [4]. Through complex NPC behaviors, advanced dialogue systems, photorealistic game worlds, and adaptive environments, AI has taken games to new levels of immersion and realism [2], [3], [4]. These advancements have contributed to the improvement of player experiences in ways that would not have been possible without the incorporation of AI [2].
The Future of AI in Games
Since the COVID-19 pandemic, there has been a large surge in the number of people who game [4]. In 2022, Statista estimated the number of gamers worldwide to be close to 2.5 billion [15]. The gaming community is expected to continue growing [4], [15]. By 2027, Statista predicts 64% of US residents will be classified as gamers [4]. AI in games is largely expected to play a role in meeting the demands of this expanding market, and even has the power to redefine the industry [2], [4].
Though their behaviors have come far in recent years, there is still potential for AI NPCs to grow [5]. According to a recent study conducted by Inworld AI, 99% of gamers felt that sophisticated AI NPCs would further improve gameplay [5]. Thanks to advancements in machine learning and natural language processing, advanced AI NPCs are expected to reach new levels of realism and dynamism in the future [5]. Advanced AI NPCs will be able to carry more natural conversations through technologies such as LLMs [5]. They will also display a greater sense of awareness and consciousness [5]. This has the potential to revolutionize games, providing players with more complex and enriched experiences [5].
Figure 6. Example from Sony’s AI PlayStation Assistant Patent. [16]
Further, AI has the potential to make gaming more accessible for players who struggle with a game’s controls or difficulty level [6]. In April of 2024, Sony filed a patent for a conversational AI assistant for PlayStation gaming (Figure 6) [6]. Players would have the ability to ask the assistant for help in real time [6]. Sony also filed a patent for an AI-driven autoplay feature [6]. This feature would allow players to skip difficult game segments that they are unable to complete themselves [6].
Conclusion
Since the 1950s, AI and games have been evolving alongside each other, with AI becoming a valuable part of the gaming experience [2]. Over the past 75 years, AI has become significantly more incorporated into both gameplay and the development process of games [2]. As its potential grows, AI is predicted to have a greater impact on games, benefiting everyone involved in games and enhancing the entire industry [2].
For gamers, developers, and even those simply curious about the future of AI, understanding the relationship between games and AI is becoming more relevant than ever. The innovations driven by their connection will continue to shape the entertainment we consume, as well as the way we create new virtual worlds.
References:
[1] IBM, “The games that helped AI evolve | IBM,” IBM. https://www.ibm.com/history/early-games [Accessed February 16, 2025].
[2] Cassiopeia, “The Evolution of AI in Gaming: Shaping the Future of Play,” Medium. https://cassiopeiaservicesltd.medium.com/the-evolution-of-ai-in-gaming-shaping-the-future-of-play-1f4ccb9a2596 [Accessed February 18, 2025].
[3] Columbia Engineering, “AI in video games,” CU-CAI. https://ai.engineering.columbia.edu/ai-applications/ai-video-games/ [Accessed February 15, 2025].
[4] C. Bharadwaj, “How AI in Gaming is Redefining the Future of the Industry,” Appinventiv. https://appinventiv.com/blog/ai-in-gaming [Accessed February 15, 2025].
[5] K. Gibbs, “Council Post: How The Games Industry Can Leverage Advances In AI To Revolutionize NPCs,” Forbes. https://www.forbes.com/councils/forbestechcouncil/2023/03/15/how-the-games-industry-can-leverage-advances-in-ai-to-revolutionize-npcs/ [Accessed February 17, 2025].
[6] P. Ferdinand, “Sony Working on AI Assistant for PlayStation Gaming,” Gamerant.com. https://gamerant.com/sony-playstation-gaming-ai-assistant-patent/ [Accessed February 18, 2025].
[7] D. Yao, “25 years ago today: How Deep Blue vs. Kasparov changed AI forever,” AI Business. https://aibusiness.com/ml/25-years-ago-today-how-deep-blue-vs-kasparov-changed-ai-forever [Accessed February 16, 2025].
[8] IBM, “Watson, Jeopardy! champion | IBM,” IBM. https://www.ibm.com/history/watson-jeopardy [Accessed February 16, 2025].
[9] “A Comparison of Chess and Go | British Go Association,” www.britgo.org. https://www.britgo.org/learners/chessgo.html [Accessed April 20, 2025].
[10] Skooldays, “Pong Game,” Skooldays.com. https://www.skooldays.com/blog/pong/ [Accessed March 3, 2025].
[11] D. Klosowski, “The Art of the Enemy: AI in Video Games”, Medium. https://medium.com/@danielklosowski/the-art-of-the-enemy-ai-in-video-games-ab565eb97f8e [Accessed March 2, 2025].
[12] G. Mason, “The Secret 80s History Of Pac-Man,” Antstream. https://www.antstream.com/post/the-secret-80s-history-of-pac-man [Accessed March 3, 2025].
[13] D. Smith, “‘Red Dead Redemption 2’ player follows NPCs for a full day,” Business Insider. https://www.businessinsider.com/red-dead-redemption-2-following-npcs-for-full-day-video-2018-11#jameson-wakes-up-in-a-bar-at-645-am-and-heads-to-work-at-his-construction-job-9 [Accessed March 3, 2025].
[14] J. Hruska, “New Intel AI Turns GTA V Into a Photorealistic Game,” ExtremeTech. https://www.extremetech.com/gaming/322803-new-intel-ai-turns-gta-v-into-a-photorealistic-game [Accessed March 3, 2025].
[15] F. Zandt, “Infographic: The World’s Biggest Gaming Nations,” Statista Infographics. https://www.statista.com/chart/28244/countries-with-the-highest-estimated-user-penetration-in-the-video-game-segment/ [Accessed February 18, 2025].
[16] “WIPO – Search International and National Patent Collections,” Wipo.int. https://patentscope.wipo.int/search/en/detail.jsf?docId=WO2019183343&tab=PCTBIBLIO [Accessed March 3, 2025].
