Artificial Intelligence (AI) has been a major buzzword for some years now. Ever increasing applications of AI see the light, from targeted advertisement to precision medicine. This also happens to be the case in the safe gambling area where AI has come to stay. However, AI is not a cure-all unless applied intelligently and exploiting the knowledge and experience of domain experts. In order to get the full benefit of AI, we need to know exactly what we are looking for and how to find it in the vast amounts of data.
The most basic application of AI is to simply search for specific risk markers and patterns. Patterns can certainly usually be identified, but are these patterns useful or predictive?
The use of AI in the shape of unsupervised learning or outlier detection is questionable when it comes to the identification of at-risk gamblers. When looking to identify this cohort of gamblers it is the severity of the behavior not just the behavior that needs to be identified.
Unsupervised learning AI offers no guarantees that the identified patterns correlate with the severity of the behavior. This occurs as the detection type is limited to identification of differences alone, as the applied algorithms applied receive no training.
A quick example that highlights the pitfalls and blunt nature of an untrained type of AI usage:
Two people spend €4,000 gambling during the course of a month, a value that exceeds the common expenditure for customers with a particular operator. The AI algorithm therefore considers both to be outliers.
One of them gambles regularly while the other has spurts of gambling. One of them stops at random, the other one chases losses. One of them plays in the evening, the other one during the night. One of them makes €40,000 a month, the other one makes €3,000. With risk markers set to alert at less than €4,000 both gamblers would fall into the category of possible problem gamblers, but when we decipher their gambling behavior, only one of them seems to have a problem.
The value of applying an AI solution ultimately depends on having an appropriate target. Without a target or goal, the outcome diminishes in value. An example from a different realm highlights this. Using AI to predict stock market prices has a clear target; that is price. Conversely, if we use AI in the stock market to simply find patterns of behavior it may only find patterns of selling, buying or both. While, within this example, not paying attention to the price pattern does not tell us very much. The application of a target in AI provides the value as it guides as to where we are headed and what we are looking for.
In the safe gambling area it has become popular to use self-exclusion as the target for AI predictions. As stated in the PWC Remote Gambling Research report, however, 80% of those who perceive themselves as problem gamblers have never used a self-exclusion tool. Furthermore, only 31% of gamblers who have self-excluded in the past define themselves as problem gamblers. Thus, self-exclusion is not a reliable target as there is no direct correlation between self-exclusion and problem gambling.
So what to do? For AI to provide relevant and useful data in the safe gambling space further thought is required. As discussed, setting externally defined limits doesn’t suffice. We need to find a deeper understanding of the mechanisms in gambling. This is where the human touch and human experience comes into the picture.
Gambling experts and psychologists understand the complexity of gambling behavior and the human psyche. With a scientific foundation and hundreds of hours spent assessing and treating problem gamblers, they are ideally equipped to spot emerging signs of at-risk behavior when presented with an individual’s gambling ‘footprint’.
Synergizing knowledge from brain imaging studies, classical behavioral analyses and their own practical experience, expert psychologists can quickly identify any concerning patterns across combinations of player characteristics including time-dependent expenditure, risk-taking, sensitivity and reaction to wins and losses, loss chasing and game types, to mention a few.
While none of these characteristics necessarily imply at-risk gambling individually, trained experts make their assessment by considering their interactions, modulations and relative influence. The result is a highly nuanced evaluation of what is actually going on in a player’s mind as the gambling trajectory unfolds, integrating elements from behavior to neurons.
Thus by training AI with thousands of deep expert assessments, it is possible to create highly advanced solutions that identifies at-risk and problem gamblers with the same accuracy as human experts. Another benefit of training algorithms with expert assessments is that whenever at-risk or problem gambling patterns are detected, a concise explanation can be offered.
This is the unique approach being taken by Mindway AI’s solutions. With this unique approach, subsequent dialogue with customers becomes deeper and more informed and additionally, given the panoramic assessment, strategies to minimize harm at the individual level are easily identified. A rather blunt tool tool is taken and enhanced to become a strong, valuable and ever evolving tool for organizations.
Nevertheless, problem gamblers can be found. What it takes is expert knowledge on what characterizes problem gamblers, using neuroscience to define targets based on experience, knowledge and evidence. Combined with state-of-the-art AI, algorithms can be trained to accurately identify problem gamblers. So there is no such thing as a quick fix. The algorithm is only as smart as the brains behind it.
Lisa Beuschau is the Marketing Manager at Mindway AI, a spin-out of Aarhus University in Denmark with over 10 years of research in neuroscience, neuroimaging and problem gambling; converting research results into responsible gambling solutions.
Lisa has a background in corporate communications, both working in global and Danish companies and lecturing intercultural communication. She drives Mindway AI’s marketing strategy and plays a key role in conceptualising responsible gambling solutions. She has a strong interest in human wellbeing in juggling modern life and is dedicated to creating a safe zone in which gambling is fun, not harmful. She also has experience with start-ups, having started two businesses on her own.