A research team from Johns Hopkins University in the United States has crafted a cutting - edge generative artificial intelligence tool, dubbed the "Traffic Safety Copilot." This remarkable tool is adept at precisely forecasting traffic accident risks. Drawing on the power of large language model technology, it meticulously examines data from over 66,000 traffic accidents. It intelligently evaluates both individual and combined risk factors, and then offers a "confidence score." This score is a crucial step in tackling the long - standing issue of the lack of transparency in AI decision - making processes. Through in - depth model analysis, it has been revealed that in Maryland, USA, accidents stemming from drunk driving and speeding are three times more common compared to those caused by other factors. This Traffic Safety Copilot boasts authentic predictive abilities. It can swiftly adapt to new situations and issue accurate warnings. Moreover, it has the capacity to refine its model by incorporating supplementary data, thereby catering to the diverse traffic management requirements of different regions.