A research team from Johns Hopkins University in the United States has developed a generative artificial intelligence tool called "Traffic Safety Co pilot" that can accurately predict the risk of traffic accidents. The relevant research results were published in the latest issue of the journal Nature Communications. The occurrence of traffic accidents is often intertwined with multiple repetitive and complex factors such as weather, traffic flow, road design, and driving behavior. This tool uses big language modeling technology to analyze over 66000 traffic accident data, including road conditions, blood alcohol concentration values, satellite and field images, thereby achieving intelligent analysis and judgment of individual and composite risk factors. The team stated that the tool not only provides predictions, but also synchronously provides a "confidence score" to visually present the reliability of the prediction results. This feature effectively solves the problem of artificial intelligence (AI) decision-making being like a "black box", removing key obstacles for AI applications in high-risk areas. Data shows that the death toll on highways in Maryland, USA, has risen from 466 in 2013 to 621 in 2023. Model analysis shows that the number of accidents caused by drunk driving and speeding is three times higher than other factors. Unlike commonly used machine learning techniques that can only analyze based on historical data, this tool has true predictive capabilities. Even in the face of new situations that have not appeared in the training samples, it can generate accurate warnings. What is even more worth looking forward to is that this tool can continuously optimize the prediction model by supplementing data, and flexibly adapt to the traffic management needs of different regions. (New Society)