INSIGHT ARTIFICIAL INTELLIGENCE
Francesco, how is machine learning being utilised to recognise and address problematic gambling? And how does the analytics platform, BetBuddy, predict those players at risk of harm?
Correctly identifying those players who might be at risk of experiencing gambling related harm is one of the biggest challenges in the whole gambling industry. Even if a minority of players can be harmed by gambling, it is critical – and an ethical imperative – to spot them at the earliest possible stage, so to maximise the chances of mitigating that risk through an effective intervention.
Traditionally, gambling operators used simple scorecards based on basic thresholds, such as money deposited or staked, and time spent playing. Over time these have become more complex and added more and more measures. But this one-size-fits-all approach has a fundamental flaw: it tends to consider all players the same, while, as we know, the human behaviour is much more complex and diverse.
Machine learning offers an invaluable opportunity to try to ‘read’ people’s behaviours. For example, in 2019, an AI poker bot developed by Facebook and Carnegie Mellon, beat some of the world’s best professional players. Using AI, the software learnt to understand when the human players were bluffing and outsmarted them by bluffing itself.
BetBuddy, Playtech’s proprietary AI software, aims indeed at ‘reading’ the players’ behaviours to spot the ones at risk at an early stage. It relies on machine learning models, that process, for every player, up to 70 behavioural markers to pattern match them with those who are known to have experienced harm (e.g., regular players who eventually decide to self-exclude or declare they suffer harm from gambling). Te models are supplemented with ‘expert rules’ to further improve the data-driven approach, and, most importantly, are fully explainable.
Te explainability layer of the models, by listing the individual markers that drive the risk, allows the operator to carry out a personalised intervention, for instance through automated in-play messages based on each behavioural profile. In our trials, AI-driven interventions proved to be up to 21 times more effective than previous blanket email responsible gambling campaigns.
How can machine learning help achieve the sustainable development of the gambling industry long term?
Machine learning enables safer gambling P98 WIRE / PULSE / INSIGHT / REPORTS
Francesco Rodano Chief Policy Officer, Playtech
approach focused on the individual players, as opposed to the traditional one, generically aimed at the whole gamblers’ population. Te latter has characterised our industry for the last 25 years and is only partially effective in tackling the problem gambling issue.
“Machine learning offers an
invaluable opportunity to try to ‘read’ people’s behaviours. For example, in 2019, an AI poker bot developed by Facebook and Carnegie Mellon, beat some of the world’s best
professional players. Using AI, the software learnt to
understand when the human players were bluffing and outsmarted them by bluffing itself.”
Francesco Rodano
Tis lack of effectiveness, especially in mature gambling markets, has been scrutinised by media and policymakers, who are now taking action by introducing increasing restrictions, such as maximum stakes or spending limits.
Te impact of these simplistic, one-size-fits-all measures on public consensus is always positive, and superficially at least they give the impression of greater restrictions being imposed on the gambling industry. However, their effectiveness in reducing harm is often limited, leading governments to introduce further restrictions, triggering a spiralling process that affects the industry, but never solves the problem.
Tat is why this area is so relevant for us. If, as an industry, we just cope with the new restrictions being introduced, without trying to break this spiral, in the medium-long term the impact on revenue could be significant. Instead, we have been researching (and advocating) on a more personalised, and effective, way to protect vulnerable players.
What are some of the disadvantages of machine learning in the context of minimising gambling related harm?
“Despite decades of research, we still lack a unique and
universally accepted definition of problem gambling, making the model training more challenging. BetBuddy has tried to solve this by using players who have self-
excluded from gambling as a proxy for harm, which is the best training set currently
available, even though some people self-exclude for
reasons other than harm from gambling.”
Francesco Rodano
In gambling, training a machine learning model meant to spot people on the path to experiencing gambling-related harm requires a set of known problem gamblers.
Despite decades of research, we still lack a unique and universally accepted definition of problem gambling, making the model training more challenging. BetBuddy has tried to solve this by using players who have self-excluded from gambling as a proxy for harm, which is the best training set currently available, even though some people self-exclude for reasons other than harm from gambling.
Players diagnosed as suffering harm from gambling by clinicians would constitute a better training set, but the number of individuals assessed this way is relatively low, and in any case their details, for obvious privacy reasons, are not known by the operators.
Tere are other ways to improve the training, for instance by considering information sources in the wider big data environment, like behaviour
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84 |
Page 85 |
Page 86 |
Page 87 |
Page 88 |
Page 89 |
Page 90 |
Page 91 |
Page 92 |
Page 93 |
Page 94 |
Page 95 |
Page 96 |
Page 97 |
Page 98 |
Page 99 |
Page 100 |
Page 101 |
Page 102 |
Page 103 |
Page 104 |
Page 105 |
Page 106 |
Page 107 |
Page 108 |
Page 109 |
Page 110 |
Page 111 |
Page 112 |
Page 113 |
Page 114 |
Page 115 |
Page 116 |
Page 117 |
Page 118 |
Page 119 |
Page 120 |
Page 121 |
Page 122 |
Page 123 |
Page 124 |
Page 125 |
Page 126 |
Page 127 |
Page 128 |
Page 129 |
Page 130 |
Page 131 |
Page 132 |
Page 133 |
Page 134 |
Page 135 |
Page 136 |
Page 137 |
Page 138 |
Page 139 |
Page 140 |
Page 141 |
Page 142 |
Page 143 |
Page 144 |
Page 145 |
Page 146 |
Page 147 |
Page 148 |
Page 149 |
Page 150 |
Page 151 |
Page 152 |
Page 153 |
Page 154
