AI-ming high: the integration of AI into gaming

Since gaming’s inception in 1958, when physicist William Higinbotham created what is thought to be the first video game, the landscape of the industry has undergone a remarkable transformation as many break-through technologies change the way in which we interact with the medium. The most recent, and perhaps most significant, change to gaming is the integration of artificial intelligence (“AI”) technologies. From more dynamic and responsive non-player characters (“NPCs”) to procedurally generated content and immersive experiences, AI is quickly becoming a cornerstone of modern gaming. This article will explore the current applications of AI in gaming, consider its future potential and address some of the key accompanying legal and commercial considerations. 

The size of the gaming industry

As noted above, as the technology used to develop games has advanced, so has the size of the industry. Whilst video games quickly took off during the golden age of arcade video games from the late 1970s to early 1980s, this success pales in comparison to the now estimated $200+ billion behemoth that the industry is today. To put that into perspective, the gaming industry is worth more than the film and music industry combined. This growth is expected to continue, with some estimates suggesting that the industry will reach a valuation of $470 billion by 2030.

As such, for those operating within the sector, great care should be taken to identify new trends in technology to remain attractive in a fiercely competitive market. As we discuss below, the integration of AI into game offerings is quickly becoming the new standard and, as such, players and developers alike should be aware of its uses and the associated legal and commercial issues.

Current uses of AI in gaming

The incorporation of AI into the industry has brought about significant advancements, transforming the way we, as players, connect, experience, and interact with virtual worlds. Whilst there are numerous examples of AI being used in this space, it is prudent to separate the implementation of AI into two distinct categories: the player experience and the development process.

Enhancing the player experience 

NPCs - The use of artificial intelligence to create NPCs, computer-controlled characters that populate game worlds, is not a novel concept. However, through the deployment of more advanced AI, NPCs can exhibit excellent intelligent decision-making, adapt to changing game conditions and interact with players and their surroundings. For example, Middle Earth: Shadow of War showcases more advanced NPCs through the use of AI integration. In this game, NPCs can act independently of the player, resulting in the NPCs making their own decisions including betraying or challenging the player. Furthermore, if you defeat an enemy NPC it can result in the character seeking revenge, making future encounters increasingly personalised and memorable. 

Natural language processing - AI-powered natural language processing has paved the way for more interactive and immersive gaming experiences. New games being sold increasingly feature AI-driven chatbots or voice recognition systems that enable natural language interactions with NPCs. Players can have conversations, ask questions and receive relevant responses within the game, enhancing the sense of realism and engagement. Star Trek: Bridge Crew allows players to use voice commands to communicate with crew and execute various ship functions, whilst West of Loathing incorporates this processing technology to create witty and humorous quest dialogues, responding to players' language and choices in an entertaining manner. 

Developers’ point of view 

Automated testing - AI can perform stress tests to assess a game’s stability and responsiveness under heavy loads by simulating different hardware configurations and network conditions to ensure the game performs optimally across a range of player setups. Tools can be similarly used to imitate player interactions within the game. This includes playtesting various scenarios, exploring different levels and interacting with game mechanics; bots can run thousands of tests in a short amount of time. TestBench, GameBench and PlaytestCloud all offer automated testing and performance analytics capabilities, aiding developers in delivering high-quality games more efficiently in both time and cost. 

Cheat detection - Cheating has always been an undesirable aspect of the gaming experience that often damages player experience significantly. Whilst developers have always tried to remove cheats, it has often been hard to detect and, subsequently, slow to be resolved. The introduction of AI to analyse player behaviour and gameplay patterns has significantly increased game developers’ ability to identify suspicious activities that may indicate cheating or hacking. By comparing individual player behaviour to baseline data, AI systems can flag anomalies such as unusually high accuracy or inhuman reaction times, to raise potential cheating alerts. Riot Games uses an advanced anti-cheat system called Vanguard for their popular game Valorant. Vanguard utilises behavioural analysis and machine learning algorithms to track players’ in-game actions and identify suspicious behaviour. It can detect cheating patterns like aim bot usage, wall hacks and other unfair advantages to issue bans accordingly. In Blizzard Entertainment’s Overwatch, players can report suspected cheats and hackers. The reports are analysed using AI-driven algorithms to cross-reference gameplay data and determine the validity of the claims. If the claims are confirmed, appropriate penalties are applied to the reported players. 

Content generation - Generative AI allows developers to procedurally generate aspects of the game, such as levels, maps, terrain and puzzles. This approach allows for infinite possibilities and reduces the need for manual content creation, ultimately allowing players to enjoy games with extended life cycles for years. No Man's Sky has utilised procedural generation to create a virtual galaxy with quintillions of planets, each with a unique weather, terrain and ecosystem. 

Bug detection - AI algorithms can also help detect and identify bugs and glitches in code more accurately. This section of game development is traditionally carried out in the Quality Assurance (“QA”) phase, but it is extremely costly and time intensive as it is carried out by manual testers (and there is no guarantee that they will be able to capture all potential bugs and glitches). By analysing vast amounts of data and patterns, AI can pinpoint potential issues that might have been overlooked during manual testing. This ensures that developers can address bugs before the game reaches players, reducing post-release patches and improving the overall gaming experience. Current examples include Ubisoft’s Commit Assistant and Meta’s SapFix. 

Where next?

As technology continues to advance, AI holds the promise of revolutionising various aspects of gaming, pushing the boundaries of what is possible and enhancing player experiences in unprecedented ways. 

Player modelling - AI algorithms have the potential to analyse player behaviour, preferences, and skill levels to create personalised gaming experiences. By understanding individual players’ tendencies, games can dynamically adjust difficulty levels, suggest appropriate challenges and tailor content to suit their preferences. This level of personalisation ensures that each player has a unique and engaging experience, catering to their skill level and play style. 

Game development - Game developers and designers could streamline the development process by using generative AI to create the initial basis for an open-world game, expediting the game development timeline overall. AI can also be employed to support post-launch development. By analysing vast amounts of player data and feedback, AI algorithms can identify patterns and optimise game mechanics. This can result in more balanced gameplay, improved pacing, and enhanced player satisfaction. For example, AI can analyse player interactions in a first-person shooter game to determine the most enjoyable or effective weapon balance, leading to a more engaging and fair experience for all players. 

Virtual assistants - AI-powered virtual companions within games can provide intelligent and context-aware assistance, tutorials, and hints to help players learn and improve their skills. These virtual assistants can adapt to the player’s progress and provide tailored guidance, enhancing the learning curve and overall enjoyment of the game. 

Immersion and cost efficiency - AI can assist human creativity and reduce costs in several areas. AI-powered text-to-speech technology has advanced significantly in recent years. Instead of hiring voice actors for every character in a game, developers can use AI-generated voices for certain non-critical characters or repetitive lines. AI-driven voice translation and localisation tools can also help by automatically translating the script and generating voiceovers in multiple languages. Whilst not released by the developer officially, an example of this is in World of Warcraft. A readily-available, player-made modification of the game (commonly referred to as a “mod”) called “GPT Voice Over” uses AI to scan thousands of lines of dialogue and automatically generates voice overs, something not commercially viable previously, but has helped to significantly improve immersion. It is important to note that the video game modding community often act as a barometer for changes to come. Natural language processing algorithms can assist in writing and optimising dialogue lines. This can speed up the scriptwriting process and ensure that dialogue fits the characters and context more effectively. Finally, as previously discussed AI-driven testing can help test the game for errors more efficiently in both time and cost. 

Practical issues and legal considerations of AI in gaming

While the integration of AI in gaming brings numerous benefits, it also raises important legal and ethical considerations. As the prevalence of AI increases, it is crucial to address these issues to ensure responsible and ethical use of the technology. 

Data Privacy - One significant concern is data privacy. AI-powered games often collect and analyse vast amounts of player data to personalise experiences and improve game mechanics. It is essential to establish clear guidelines for data collection, storage, and usage to protect players' privacy and ensure transparent data practices. One way of achieving this is by clearly stating in each games’ End User Licence Agreement (commonly referred to as a “EULA”) what data is being collected and processed.

Data Bias - Algorithmic bias is the presence of unfair and discriminatory outcomes caused by the underlying algorithms and data used in game development and operation, such as test data. Whilst AI has a proven track record of being significantly beneficial, there are concerns that bias can arise from some algorithms and, as such, developers should consistently monitor for any biases that may arise.

Skill and ranking bias - Similar to the issue above, bias in skill-based matchmaking (“SBMM”) algorithms can impact the gaming experience for players. When SBMM is implemented without careful consideration, it can result in unfair matchups, pitting highly skilled players against novices. This not only creates frustration but also hinders players' opportunities for growth and improvement. That said, an overly strict SBMM approach could lead to longer waiting times, creating player drop-off and a less engaged player base. Striking the right balance will be crucial in ensuring that AI is implemented effectively within gaming. 

Monetisation bias - Whilst AI-driven monetisation strategies have become standard in many free-to-play games (including games like Candy Crush), there are risks associated with the misuse of these algorithms. Monetisation bias can lead to potentially harmful practices, as AI can analyse player data to identify potential high-spenders and target them with personalised offers. This targeted approach could lead to vulnerable players being encouraged to spend more than they can afford or engaging in compulsive buying behaviour. Without monitoring this risk, companies may attract negative press articles covering extreme examples and risk the overall gaming environment becoming less safe. 

Intellectual property - Intellectual property rights are another area of concern. As AI becomes more involved in generating game content, questions may arise regarding the material used to train the AI models, potentially giving rise to disputes over the ownership of the AI output and copyright. Developers must navigate the legal landscape to clarify the ownership of AI-generated content and ensure proper authorisations, attributions and licences are obtained. 

Consumer protection - Ensuring consumer protection is crucial. As AI algorithms become more sophisticated, the need for transparency regarding AI-driven mechanics, such as dynamic difficulty adjustment or microtransactions, will increase. Players should have clear information and control over these systems to maintain user autonomy within gaming. 

Conclusion

In conclusion, AI’s integration into gaming has brought about significant advancements, enhancing gameplay mechanics, creating immersive experiences, and enabling intelligent behaviours. Its potential for future applications is even more promising, with advanced player modelling, improved game design and virtual assistant functions. However, it is vital to address the legal and ethical concerns, such as data privacy, algorithmic bias, intellectual property rights and consumer protection. By navigating these challenges responsibly, we can ensure that AI continues to shape the future of gaming in a positive and inclusive manner, delivering captivating experiences for players worldwide. 

At CMS we are not only expert legal advisers; we are enthusiastic gamers (and in some cases, game creators) ourselves. Please do get in touch if you have any questions.

Co-authored by Carter Rich, Jure Tus, and David Zeffman.

The authors of this article would also like to thank Cameron Finlay-Hylton, a CMS Academy Intern, for his help with writing this article.