AI and gambling – big opportunities but big risks too?
8 October, 2025
Keeping Innovation Investable: Why Simplicity Wins in Gaming Tech
22 October, 2025
Winna Media Opinions & Editorial
14 October, 2025
Let the buyer beware – using AI to evaluate gambling industry assets
Let the buyer beware – or caveat emptor to use the original Latin wording – remains a fundamental principle for any purchaser.
Basically, if you’re buying something, you can’t get recompensed if what you’ve bought isn’t fit for purpose, unless the seller concealed defects or misrepresented themselves.
But does that principle still apply if Artificial Intelligence (AI) was used to value a casino, gaming equipment or an online platform?
AI can process data and surface the underlying values quicker than humans. But, on the downside, that depends on the data being accurate, up to date and agreed on by all the parties involved. The exercise will also be affected by the resources available, like technical experts, to assess the information, and whether or not everyone involved has deep enough pockets to pay the costs involved.
AI evaluations – the upside
Let’s assume that the data at stake is agreed – e.g. revenue streams, customer behaviour measured over an agreed time period and historic market/regulatory trends. In that case, AI can process the data using agreed criteria faster and more accurately than it could be analysed manually.
Since there’s much less need for comprehensive manual audits, that saves time and administrative costs.
Using machine learning, models can be constructed to forecast revenues and customer churn rates based on real-time and historical data. Data sets can identify and surface detailed information on user behaviour, customer loyalty metrics and individual game play. All of this can be used to create accurate and reliable asset valuations with risk-adjusted projections depending on the volatility of the individual market.
These calculations can also highlight potential liabilities early, like the prevalence of irregular betting patterns, which could signal non-compliance with the regulatory framework. They are also particularly important for companies managing multiple gambling industry assets that need to look at their portfolios in real time.
It is also worth reiterating an earlier point that if the data is agreed by all sides, objective benchmarking that includes standardised criteria can root out human biases that transcend multiple jurisdictions and asset classes. The result is that it is much more straightforward to create like-for-like comparisons.
The devil’s in the detail
Unfortunately, the numbers may not match, or the conclusions drawn from trend analysis can be interpreted very differently by the groups involved. The performance of AI depends on the reliability, cleanliness and breadth of the data it sees. And these factors will vary according to local privacy laws, overall regulation and its varied enforcement in different jurisdictions.
So, it’s not only critical that operators share internal numbers around player retention, for example, but that the would-be buyers can see the player data they need while still being on the right side of personal information protection regimes. Given the ability of cyber-criminals to access confidential information, there are already massive concerns about protecting personal privacy, including not releasing anything without the individual’s consent.
Another issue is ensuring that ‘clean’ data is used that has had any inherent biases removed. That’s essential in ensuring that the value and/or limitations of the data are commonly understood and agreed upon.
The resources each side can commit also needs to be a factor in AI-led number crunching because developing and maintaining systems requires expertise and budget.
We still need to beware
Any observer of the ebbs and flows of Thailand’s attempts to legalise casinos over the past couple of years will agree that it is impossible to predict and accurately factor in the impact of outside events like economic downturns, faction-led politics and shifts in public opinion. Therefore, forecasts based on AI need to remember it can’t predict the future – at least not yet. That means involving human beings who can factor in the impact of the unknown or the unlikely.
We know that AI can transform how assets are measured and valued quickly and precisely. Yet unless the data is agreed, bias-free, up-to-date and accurate, the potential for argument and conflict is high. For the time being, a hybrid approach whereby humans are supported by AI and where the implications are assessed in context by experts, remains the most effective way forward.
Owen Hughes
Owen Hughes has more than two decades-worth of experience working in corporate communications and journalism in the Asia-Pacific. He's written news articles, features and analysis for a wide variety of international and regional print, video and digital platforms, as well as for internal company audiences.
