How AI and DS fits into Games24X7’s business needs

What are the components/ingredients that drive Games24x7’s Artificial Intelligence and Data Science (AI & DS) strategy across products and business verticals? (What is the ecosystem that will make it successful?) Please elaborate on each of these components.
A technology company with ‘science of gaming’ as our core philosophy, we have always worked towards understanding our players and their interests through rigorous data analysis and application of machine learning algorithms to provide them the best game play experiences.

AI and DS are integrated into the DNA of Games24x7 and aligns to the core of our business. In the last 4years, our AI and DS team has grown 8x to become a strong team across multiple verticals and specialties through strong collaboration between engineering, product & business. Forming the backbone of what we do, our AI components include:

• Game AI: game action prediction, skill assessment, game strategies, upskilling

• Process AI: Experiments, marketing customer service, risk & fraud

• Content AI: player journey optimization, prediction and diagnosis, recommendation, and personalization

Our AI and DS components when combined with various functions work towards fulfilling our business requirements. Which then enables us to bring various solutions for our products to ensure responsible gameplay for the players. 

What steps is Games24x7 taking to make these elements a reality/life?
The gaming industry is seeing an unprecedented growth and is challenging the traditional forms of entertainment. India is also showing a huge potential of becoming an online skill gaming power in the world by growing at a rate of 30% CAGR. This is transforming the online skill gaming space and bringing in newer opportunities and investments for the sector and the organization to create a strong foothold in the global market.

This is a huge opportunity for skill gaming players in the Indian market and we are ensuring that we leverage this opportunity by understanding the customer base as early as we can in their lifecycle. Our team of data scientists along with the tech team are using behavioural science, tech, & AI to understand and personalise game playing experiences for our users. Our increasing user base which is currently 100Mn+, is a huge testament to the fact that this is working.

What are the challenges faced to bring these components to life?
The key challenge is to align the work that we do as AI & DS with the key business needs. Unless it is aligned it will never get to see the light of day. Second and most important thing is to understand the product ecosystem. Learning about the product journey for users and how the different components interact and where this module fits in while trying to meet the business objective is very important. AI & Data Science cannot fit within a silo where you are given a problem and you are solving it by coming up with a model. To make it to the end of the road, it definitely needs to align with the company’s business needs. There is a need to understand the various components of the product on how they interact and operate in general for it to function well.

Finally, even if there is the understanding, there is a need to know more about the system details on how people from each vertical will incorporate or integrate to the start of the product ecosystem for each function to work efficiently. Hence, these are the challenges and ensuring that they all align to each other is important in order to bring the components to life.

What are your recommendations / best practices to face these challenges and overcome them?
Its two fold approach: One is the alignment between various teams such as product, business, data science and engineering teams on a project. Creating a project team with people from each vertical who are accountable and responsible for the project will really help in completing it faster and better, instead of working in silos.

Another is open-mindedness across stakeholders including data scientists. Even if a certain individual from a certain vertical feels that they are good at one aspect of the problem and should stick to doing that, it won’t help them in understanding the entire function of the project. The curiosity and willingness to learn other aspects of the project and work along with others on how to solve it, will automatically enable them to know the business context of the project and align smoothly with the business requirements.

How do you see the overall socio-political environment in which the gaming industry in India operates? How do you positively or negatively get impacted?
Being one of the fastest growing sectors in the country, what is happening in gaming is typical of what we have witnessed in other tech sectors, where the sector developed faster than the regulation (e-commerce, urban mobility, food delivery and OTT platforms). 

We are already going ahead and educating governments and regulators through the self-regulatory bodies that we have built which is E-gaming Federation (EGF) which mandates that its members have checks in place to ensure that players don’t go overboard. The understanding of how the sector functions and what is promotes are still unclear, mainly because of  confusions between a game of skill and game of chance. This has impacted the gaming industry at large, due to services being shut. While there were a few shortcomings in various states, in the fullness of time,  some states were able to realise the importance of this sector which has enabled it to get back on track.

What are the kind of data science and AI tools (off the shelf and open source) that are making waves and how has been your experience using them at Games24X7?
At Games 24×7, we use art to create art. The beautiful and interesting outcome is purely a result of our detailed and hefty framework. The tools that enable us to bring awesome games for our users are:

• Responsible Game Play (RGP) – Our responsible game play has various components that help with our work functions and solution modelling for our players. These include:

• • Data ETL and Feature Preparation Pipeline – It dynamically captures and models a user’s behavior using multiple data sources – gameplay, amount and sources of the transaction, clickstream, exogenous data, and so on. For e.g.: to ensure responsible gameplay – we select features along three dimensions of immoderation – Time, money, and urgency to play. 
  • Explainable AI – This allows us to take actions based on the results of a game/ product model. It is significant in the functioning of any game play model that we develop.
  • Domain Expert Based Rule Engine – Map users and their behaviour patterns over the dimensions of time, money, and urge to play. All players are assigned scores on these dimensions and those surpassing the thresholds on a certain number of dimensions are flagged
  • Adversarial Auto-Encoder based Pipeline – AAE is a combination of generator and discriminator network; encoder and decoder network. The two networks, when trained back-to-back, learn the complex distribution of the healthy players in the provided space. 
  • Game Play Desperation Model – We establish certain irregular behavior-driven rules to categorize this urgency to play. 
  • Local Expert – This pipeline primarily uses time series of two types of parameters – player’s wallet recharge patterns and game entry fees. These time series are studied to create new features.

• Counselling Process – We have built a reporting system that updates and reports the end-to-end journey and gameplay status of the player along with the accuracy of each of the models on a daily basis. 

• Reinforcement learning – majorly used for game play modelling which is a traditional use of AI for gaming. We also use it for simulating gameplay not in or real money gaming like RummyCircle or My11Circle, but for U Games, our casual games studio. Reinforcement Learning is also an important lever we exploit for Hyper-personalization.
• Computer Vision – Computer image representation is used to capture game states that the user is in and then computer vision techniques are applied to it. It is also used for game production with respect to art, design, content and creatives that can aid in making our game even more exciting for our users
• Cognitive Neuroscience – It enables us to understand our users better by mapping their telemetric data that is collected from their game play experiences. Hence extrapolating their nature and behavior on our platforms
• Sequential modelling – This enables us to learn the journey and evolution of users on our platforms. This is then mapped in order to determine the users’ future journey on our platform
• Procedural content generation – Used for generating content in game production, especially while creating various levels in a game 

Being in the skill gaming industry, we are in a very unique position to leverage data by understanding user behaviour, game play experiences, the choices they make and the content they consume. Hence, these tools enable us to create personalised and safe game play experience for our users.

Many states are taking policy decisions that stifle the gaming industry. Does it harm the way in which you operate (I mean if the gaming industry is regulated you might not have access to certain publicly available datasets data critical for your DS and AI strategy)
When we talk about regulatory policies, the privacy of user data has always been an important aspect for the government. However, this is not just limited to the gaming sector but to the entire internet ecosystem, be it OTT platforms or social media platforms or even browsing certain websites, user data has always been readily provided while trying to access these services. Hence, when it comes to gaming, the intention of regulation should not completely be from a user data perspective but rather the industries intent of personalising or monetising the users. Because, as platform providers it is very important for even us to have that level of responsibility and regulation. As one of the leading multi-gaming platforms, we have been doing so, to ensure the safety and privacy of our users.

As many state governments are taking policy decisions for the online gaming industry, we as an organization have already been working on self-regularisation. At Games24x7, we have been applying AI, ML and data science at an advanced state of our research to come up with certain solutions in order to provide a responsible gameplay experience to our players.

To the contrary, the government policies and regulations impacting the AI and DS function, we believe that these technologies are only going to aid the government in their decision making. In fact, we believe the government should leverage AI, ML and DS in order to map the sector, understand user behaviour, see how companies are working of creating a safe space for the players and provide proper regulatory policies.