Blog

The pandemic has hauled us into the future with unprecedented speediness. The technologies that were being talked about as those of the future – data analytics, Artificial intelligence (AI), Machine Learning (ML), and hyper-automation – are now the necessities of today’s tech world.

Their accelerated adoption has completely revolutionised the tech ecosystem in the ‘evolving normal’ across sectors such as ecommerce, fintech, healthcare, and social media.

The gaming sector, despite being relatively less mature than the others, is no different. Integration of AI in the science of gaming is no longer restricted to only modelling moves for the game, but goes further to enable a 360-degree insight into the gamer.

An online gaming platform will typically have massive data coming in, which includes in-game actions, player moves, clickstreams, and transactions to name a few.

The varied data sets can be fed into AI and ML systems to give unique information about the users. It is rather interesting, as something as simple as data on gaming moves can help create a psychological imprint of the user at that moment, based on his impulsive reactions and response to a situation in the game (which we call game state.)

We could perhaps draw a parallel to social media, where user-generated content is explicit and can tell a lot about the user. It will be surprising for many to know that the implicit nature and the impulsiveness of user responses in-game can be used to create a more authentic identification of the phycological imprint.

Also, because of the clickstream data, it is possible to understand and even personalise the users’ journey in the platform.

While it is similar to personalisation and recommendation systems in consumption-based entertainment platforms such as OTTs, it is also significantly different as gaming platforms are highly interactive and the consumption of the user is dependent on the outcome and experience on the platform.

Let’s say, for example, deciding to continue to play a game as a response to a win or loss is starkly different from purchasing an item over ecommerce website or listening to the music video of choice.

In this scenario, the end outcome is a direct result of user action and in cases when it is not favorable, say in case of losses during online gaming, the user might want to recover by playing more.

This sets an important use case for AI and data science to leverage the 360-degree user insights gathered through multi-dimensional data to calibrate and personalise user experience and game behaviour. This will play a huge role in ensuring fair and safe gaming environments.

It also raises important questions about responsible AI. One must be cognizant of the fact that AI technology is evolving at a speed faster than our ability to deploy it responsibly and ethically.

ALSO READ

Play and Earn: Is a decentralised gaming metaverse the future?

Issues like AI bias, model auditability, and explainability are pushing the need for regulations. Recently, companies such as GM, Nike, and Walmart announced a collaborative, cross sector Data and Trust Alliance to address and resolve algorithm bias.

Responsibility is usually used from the perspectives of fairness and explainability. Data scientists need to be vigilant and account for data bias while developing models to be fair.

It is also important to make sense of the inferences made by a model – hence explainability. Let me make it simpler to understand. If users from a certain community are the major users of an online music platform, it is likely that the data would represent the preference of that community over the others.

In an intent to personalise music recommendations, it is possible that the model learns from past user choices and gets biased by the preferences of a specific community, thus being unfair to other communities.

AI and Data analysis systems need to account for such biases. Even after removing the biases, it is quite possible that the recommendations remain un-interpreted by businesses as ML models might be unable to communicate what they have learnt and hence make skewed recommendations.

It thus becomes imperative, especially from a responsibility standpoint, to integrate a system that can explain the ML model. Modern research is actively working to come up with techniques and measures to set specific data variables important for model predictions.

From the gaming perspective, massive data generated with the interaction of the gamers with the platform can be leveraged to eventually understand the psychometrics of the gamer via innovative design and application of data science and AI/ML models. This will be essential to ensure player wellness and to prevent overindulgence.

Enhanced and extended AI systems can identify the users going in the direction of overindulgence and potentially risking themselves. This assumes all the more importance when conventional, subjective psycho-analysis remains impossible to undertake for millions of users on a platform and are this neither feasible nor scalable.

A well sequenced AI/ ML model will be imperative for establishing a healthy gaming culture, ensuring sustained user engagement while creating a safe, healthy online gaming space. This will truly enable the potential of this new-age entertainment sector, which is growing exponentially in India and globally, to be fully realised.