Researchers Develop AI Tool Predicting Personality and Mortality Based on Life Events

A new AI model called life2vec has been developed by researchers, which can predict human lifespan by analyzing sequences of life events. The tool utilizes transformer models and is trained on a dataset comprising the entire population of Denmark. While it exhibits impressive accuracy in predicting various aspects of individuals’ lives, including mortality, the researchers emphasize that it should not be applied to real people. Instead, they view it as a foundation for future research and as a means to gain insights into societal policies and regulations. By involving social scientists in the development process, the team aims to ensure a human-centered approach to AI development and prevent the tool from overshadowing the importance of human subjects. The core of the model lies in the massive dataset provided by Statistics Denmark, which contains detailed information on every Danish citizen. The researchers employed recurring life events to create a representation of a human life as a sequence of events through the transformer model approach. Published in Nature Computational Science, the study highlights the comprehensive reflection of human existence that life2vec offers compared to other models. Despite the potential benefits, the researchers emphasize the need for ethical and responsible usage of such AI tools.