AI has learned to analyze the causes of traffic accidents and predict new ones.
Specialists from Johns Hopkins University (USA) have created an artificial intelligence (AI) tool capable of predicting road accidents. The research was published in the journal Nature Communications.
The team of scientists developed a program called SafeTraffic Copilot, which was trained to analyze the causes of traffic accidents as well as accurately forecast future accidents. According to the specialists, AI will help reduce the number of accidents and casualties on American roads.
The large language models (LLM) used by the researchers were trained on text, digital values, and satellite data. The SafeTraffic Copilot model can assess both individual and combined risk factors. The program can explain whether certain events and circumstances preceding accidents are related to each other.
“With AI, we can move from rough, aggregate statistics to a precise understanding of the causes of specific collisions,” noted study author Hao Yan. SafeTraffic Copilot is continuously learning, allowing it to take into account new circumstances. The accuracy of the AI's predictions was estimated by the scientists to be 70 percent. “The number of car accidents in the USA continues to rise despite countermeasures. These are complex events influenced by many factors such as weather, traffic patterns, and driver behavior,” Yan concluded.
Leave a comment