Professor Jessica Staddon from Northeastern University is researching the efficacy of AI tools like Gemini and ChatGPT in detecting fraud, driven by a significant increase in consumer fraud complaints—over 2.6 million were reported in 2023, more than double from 2020. Staddon and her team are developing effective prompts for these AI models using language from real consumer fraud reports to help them distinguish between scams and legitimate communications.
Initial findings suggest that while these AI models show promise, they require further training. For instance, the tools sometimes detect scams based on context rather than traditional markers, and they exhibit bias toward known organizations. Staddon highlights the urgency of providing consumers with clear guidance on how to interact with AI for fraud detection, especially given the rise in scammers adapting their tactics. Building prompts with actual consumer language is intended to help the AI recognize patterns in fraudulent communications.
Source link
