AI and responsible gaming
David Sachs, CEO of TomoBox, says that artificial intelligence is now driving regulation in iGaming.
Data, not oil, is now running the world. This is evident by the lavish market valuations of data related companies like Apple, Facebook, Google and Amazon. Hence the new kingpin sifting through those huge amounts of data is Artificial Intelligence (AI).
AI, in its practical form of computer science, is a set of algorithms that can predict an outcome based on historical data. While histrionically computing power was scarce and benefited only the rich, the abundance of cloud services affordable to even the most fragile upstarts has spurred an industry wide use of AI.
Demand is meeting supply where data-driven industries are looking to capitalize on AI to improve their competitive edge. Online betting and iGaming have always been at the forefront of analytics and number crunching. Now, gaming operators are eyeing AI as a solution to a wide range of challenges ranging from player churn to fraud detection and recently gaming compliance.
Given the regulatory climate around anti-money laundering (AML) compliance and safe betting procedures, it came as no surprise when the UK Gambling Commission slapped online gaming behemoth 888 with a £7.8m fine in August 2017, and even less so more recently when online betting powerhouse Paddy Power Betfair was also rapped to the tune of £2.2m.
Each of the fines were handed down as a result of the operators’ failing to install safety measures in support of responsible gaming and preventing the use of illicit funds on their platforms.
Backed by a worldwide tightening up of illicit money trafficking by national financial intelligence units as well as global organisations such as the Financial Action Task Group (FATF) and the Egmont group, gaming lords are finally picking up on the cue to make online gaming a safer and more responsible experience.
Money laundering harbours many perils to our well-being as it is typically associated with organised crime, drugs, terror and illicit trading of wildlife. U.S. Federal agencies estimate that $300bn is laundered annually in the US alone.
Traditionally, gaming operators had little motivation to enforce responsible gaming and banking procedures such as Know Your Customer (KYC), as it added a substantial overhead to their operations and eroded profit. However, recent actions by the regulators have shown that times are a changing and both platform owners and operators should be falling in line.
Identifying gaming related infringements remains challenging for regulators, not least because of the lack of tools to sift through the huge amounts of data generated every minute by online gambling platforms.
Fortunately, the abundance of data presents a compelling opportunity, potentially helping to profile players’ behaviour and identifying patterns associated with risk, fraud, and responsible betting.
Unfortunately, current analytical tools are limited to pre-coded decision trees, aimed at very specific use cases and they falter in the face of ever-changing data.
Given the relatively low cost of cloud-based computing power, AI-based machine learning appeals as a means of alerting internet-based gaming operators with regard to players who’ve gone astray, block them and make sure these cases are well documented.
What is Machine Learning?
Machine learning (ML), a branch of computers science and artificial intelligence, is based on algorithms that can identify patterns within data.
Machine learning is at its best when working through unstructured data; going through player in-game chats that come in many languages and different variations, for example.
There are two approaches to modern machine learning. Using supervised ML, player conversations are manually marked as potentially belonging to compulsive players and later used as training data for the machine learning code.
The machine then generates a model that has a very good chance of identifying compulsive players and flagging them to be documented.
However, manual annotation of data for training purposes is costly and to some extent difficult to obtain. Enter unsupervised machine learning, which requires data but does not require the training data to be tagged.
Unsupervised classification, also known as clustering, has a tremendous value since it can point to anomalies that are very useful in identifying fraud and anti-money laundering schemes. Its core value proposition is being able to adjust as data changes with time.
For example, when detecting language that might point to compulsive players, sentences uttered by gamers such as: “This is my last pound, I’ll lose my house” or “I’m taking food from my babies”, are very different in nature but the context is the same. It indicates a person in distress.
Surprisingly enough, machines have become very astute in identifying the gambling patterns.
These might be in the form of red flagging a player which had bet with minuscule sums and just recently was placing much larger bets. In poker these might be groups of low-key players that when playing together as a group huge sums appear on the table and might point to a form of collusion.
In summary, recent developments in AI present a huge opportunity for regulators and operators to automate and enforce regulations throughout online gaming platforms.
David Shoham Sachs is the founder and CEO of Tomobox, an AI-based startup, backed by Nielsen Ventures and The Founders Group, with focus on analyzing the cognitive behavior of consumers in financial services and online gaming. His special interest is using AI to promote responsible betting and prevent money laundering.