Using AI to Power the Science of Gaming

Most people think AI in gaming is all about building the best game-playing bot with the capability to defeat the best human players and/or the ability to learn a game and reach the end goal from an initial state by surmounting obstacles on the way.

There are many reasons behind this conception. But, it is primarily formed, by the nature in which the application and research in AI and ML was conventionally projected for gaming. For many skill games such as Chess, Go, Atari and so on, and of late for games like Poker, the quest for AI/ML researchers have been towards creating the best superhuman bots. For casual games, on the other hand, the focus has been to create game characters and elements that learn automatically during the game.

Today, modern game development platforms can provide the necessary nuggets to game developers to incorporate such learning capabilities into games. Game AI researchers keep plugging in innovative learning algorithms to procedurally generate game content. In this regard, games provide an attractive, yet a non-intrusive way for researchers to design and test virtual modelling of reality and agents/bots navigating through reality, which seems to be the ongoing and ultimate aspiration of humanity.

It is obviously not possible to create game-playing bots because of regulatory reasons and moral responsibilities of the operators. The research community around human-computer interaction has given us very interesting qualitative insights on the behaviour of game users in terms of skill assessment and skill-matching, factors and designs that enhance engagement, retention and more. The recent proliferation of digital games for commercial, social and educational purposes has further necessitated a new research direction towards game intelligence and knowledge discovery. The quest here is to enable a vision of end-to-end informatics around game dynamics, game platforms and the players by consolidating orthogonal research directions of game AI, game data science and game user research.

At Games24x7, we generally look at games and players from an analytical perspective using the terabytes of multi-dimensional data generated daily from millions of games, clickstream and transactions, especially from cash game platforms such as RummyCircle and My11Circle.

All these data contain a goldmine of information about gameplay behaviour, player inclinations, intentions and preferences. Based on data-driven analysis, an evidential and experimental approach is taken towards understanding new business opportunities and enabling awesome gameplay experiences by solving those problems using an AI/ML-first approach, i.e., AI/ML becomes the heart of all user journeys and personalisation.

Towards the vision of the science of gaming, the AI & data science team dive into a plethora of multi-dimensional data while breaking away from the conventional wisdom of what AI/ML means for gaming. The problem space we deal with has a unique combination of gaming and e-commerce domains.

The AI & data science team consists of researchers from statistical and computer science backgrounds who work together to perform research in technical areas such as deep learning, reinforcement learning, stochastic games, partially observable Markov processes, image and text, to name a few. Game AI is to understand players’ skills, strategies, mistakes, etc. while playing games; Process AI enables rigorous automation in various operations that touch user journeys and Content AI leverages Game and Process AI to further understand broader behaviour, intent and inclinations of players, and accordingly provide recommendations for personalised and target journeys on our gaming platforms while enhancing the gameplay experience.

The Author

Tridib Mukherjee leads the Artificial Intelligence and Data Science team at Games24x7. He completed his PhD in Computer Science & Engineering from Arizona State University, where he specialised in AI-driven data centre optimization and body sensor networks. At Games24x7, his team is responsible for applying AI and ML techniques to automate business decision making, analysing and mining players’ skills and strategies and enhancing user journeys and gameplay experiences through personalisation.

‍