Stories
DraftKings AI Targets Losing Gamblers for Promotions, Sidesteps Problem Gambling Detection
- •A New York Times investigation reveals DraftKings uses AI and customer data to identify gamblers likely to respond to ads and promotions by betting and losing more money.
- •Former employees stated that methods were refined to direct incentives towards losing gamblers, while efforts to use similar technology to predict gambling problems were stalled or abandoned.
- •DraftKings developed a machine-learning model in 2023 to predict which customers would increase gambling and losses in response to promotions, scoring them based on gambling frequency, account balances, and typical losses.
- •Promotional offers are crucial for online gambling companies; DraftKings generated $8.7 billion in gross revenue last year and distributed approximately $3 billion in promotions.
- •Despite internal efforts to develop an AI model to spot gambling problems, the project was shut down, with DraftKings citing a lack of sufficient evidence for predictive technology, preferring its existing monitoring system.
Key Questions Answered
In 2023, DraftKings developed a machine-learning model to predict which customers would respond to promotions like free bets and bonuses by gambling and losing more money. The system scored customers using factors such as betting frequency, daily balances, and typical losses relative to their bets.
Projects exploring machine learning to detect potential gambling problems were ultimately shut down or shelved. DraftKings responsible gaming chief Lori Kalani explained that leadership decided against using predictive technology because they did not find it sufficiently evidence-based.
DraftKings monitors customers nationwide for over two dozen indicators of dangerous gambling, and when triggered, can send educational materials, questionnaires, or close accounts. Additionally, the company uses large-language models to monitor customer communications with representatives for signs of distress.
