Concerns over the use of artificial intelligence (AI) in gambling-related research have intensified after politicians and academics raised questions about an evidence review used to support a AU$20m (£10.4m) funding request for youth gambling education in Australia.

Independent Senator David Pocock, a longtime advocate for gambling reform in Australia, said he was “deeply concerned” about the report circulated by the OurFutures Institute. The MP blasted the research, telling officials it “appears to just be slop written by AI”.

He added that the document was “full of AI hallucinations, including references to studies that don’t exist and statements presented as facts that are completely false or grossly exaggerated”.

Analysis of the review by Guardian Australia found at least 21 problematic references, including broken links, citations to papers that do not exist and claims unsupported by the research cited. 

In one example, the review claimed a government inquiry found “every $1 invested in school-based prevention returns $8-10”, but that assertion doesn’t appear in the original inquiry.

The institute said an “editing tool” used to reorder references caused errors and Chief Executive Ken Wallace “deeply apologised”, adding that his company would conduct a “full, line-by-line verification” of citations.

The ChatGPT-ification of research

The episode has sharpened debate over AI’s role in gambling harm research and policy submissions. One senior figure at UK-based gambling harm prevention firm, EPIC Global Solutions, told SBC News about how their company’s efforts massively differentiated to that of the OurFutures Institute.

“EPIC’s content is designed using the latest academic research and the valuable input of those with lived experience of gambling-related harm,” said Anca-Maria Gherghel, Senior Research Manager at EPIC Global Solutions. 

“Additionally, we believe it is crucial to conduct research and evaluation on the back of our sessions, to both understand current trends and measure the effectiveness of our sessions.

“We present the findings of our independent research to leading academics, as we recognise the value of sharing valuable, verified data to help increase understanding of the potentially complex and under-reported nature of problem gambling.”

AI has been having a huge impact in the gambling sector, as with every other industry – especially ones so inherently tech driven. This has naturally led to companies and charity organisations looking at how to utilise it for responsible gambling and gambling harm prevention.

The tech’s ability to sift through vast amounts of data has come in handy for problem gambling detection, for example. Naturally it makes sense that AI would also find its way into gambling harm research.

Likewise, it is also finding use among consumers, such as for finding brands. A recent YouGov study also found that 16% of UK bettors have already used AI tools to guide gambling decisions in the past year. 

EPIC’s Gherghel continued: “Though AI has a role to play in harm detection, which we endorse, EPIC Global Solutions will only create policy, papers or presentations based on human insight to guide the process, relying on the human excellence of authentic intelligence rather than artificial intelligence. 

“We learn about emotive subject matters based on genuine human qualities such as empathy and relatability; something that cannot be accurately replicated by AI.

“We have to constantly improve and grow to ensure our interactions are as impactful they can be, and the only way to do that is through thorough research and evaluation to understand the needs of our worldwide audiences.”

AI a useful tool – when used right

Meanwhile, Rob Mabbett, Engagement Director at Better Change, said that AI can improve research efficiency but warned about risks if outputs are not rigorously checked. 

“If you take the approach that it is a useful tool and not something that will do your work for you, you will not go far wrong,” he explained.

“AI is fantastic for interpreting large data sets, we now have the capability to process information on a national, even global scale around anything we want, in this case gambling harm. 

“The key element, however, is that if you are submitting the work you are responsible for it and any research referenced or factual claims are made, they have to be fact checked.”

Questions around AI’s role in research come amid public and political scrutiny around gambling in numerous markets – including the UK and Australia, the focus markets in today’s article.

Australian MPs are pushing for changes to gambling legislation, based on the 31 recommendations of the Murphy report published in 2023. In the UK, the recommendations of the Gambling Act review are still being adopted, but some MPs are calling for another look, with a particular focus on retail gambling.

Research into gambling-realted harm plays a big role in influencing these political debates.

“One of my concerns around the use of AI is that it could reduce the quality of what is put out,” Mabbett said. “When it is an image used for a social media post and there is a spelling mistake or an extra finger, no harm done. 

“When it is a document, a proposal, or research that is seeking to influence policy to obtain large sums of money as in this case then it is a different story – there needs to be proper scrutiny. 

“I wonder out of the 1000’s of pages of submissions sent to the levy board for research, prevention and treatment what % might be “AI slop”? It’s a tool, a very powerful one granted, but still a tool.”

As stated above, the controversy comes as governments and regulators increasingly rely on academic and policy submissions to shape gambling harm prevention strategies.

These have been ramped up in recent months around the world, particularly in places such as the UK, Romania and the Netherlands. As recently as last week, Wolverhampton Council opened a public consultation to inform its Gambling Harm Reduction Strategy. 

On top of this, an academic study from researchers at the University of Sheffield found that current rules governing gambling advertising may not sufficiently limit its influence on betting behaviour, after it investigated behaviours of 18-45 year olds during the 2022 FIFA World Cup.

It could be argued that these measures are undermined when established organisations rely on the use of AI to inform their strategies to protect the public from dangerous gambling.