A groundbreaking AI system, similar to ChatGPT, has made waves in the field of predicting individual life outcomes and even the time of death. Developed by scientists from the Technical University of Denmark (DTU), the artificial intelligence model, named life2vec, was trained on the life stories and personal data of over a million people in Denmark. The study, published in the journal Nature Computational Science, revealed that the model outperformed existing systems in accurately predicting outcomes like personality traits and time of death.
The researchers utilized a dataset that included health and labor market information for 6 million Danes from 2008 to 2020. By transforming the data into words to train the language model, the AI system demonstrated an impressive accuracy rate, surpassing other models used by life insurance companies. In fact, the predictions made by the model were found to be 11 percent more accurate compared to existing AI models and methods.
To test the effectiveness of the AI model, the researchers applied it to a group of individuals aged 35 to 65, half of whom passed away between 2016 and 2020. The results showed that the model’s predictions far exceeded those of other systems in terms of accuracy. Sune Lehman, the lead author of the study from DTU, explained that the model’s ability to analyze life events and sequences played a crucial role in its success. Drawing parallels between human life and a sentence in language, Lehman emphasized that the AI system grasped the patterns in the data and showcased remarkable precision in outcomes such as personality traits and even the time of death.
Through its analysis, the AI model discovered associations that shed light on certain factors influencing life outcomes. For instance, individuals in leadership positions or with higher income were found to have a greater likelihood of survival. On the other hand, being male, skilled, or having a mental diagnosis increased the risk of an earlier death.
While the AI system showed tremendous potential in predicting outcomes and personality traits, the study’s authors expressed ethical concerns and advised against its use by life insurance companies. Dr. Lehman pointed out that the predictions made by the model could be at odds with the underlying principle of insurance, which is based on sharing the lack of knowledge regarding who may be affected by unforeseen circumstances.
Despite these concerns, the researchers believe that the model’s ability to provide precise answers opens pathways for understanding new potential mechanisms that influence life outcomes. It also paves the way for exploring personalized interventions to improve individual well-being.
In conclusion, the development of the life2vec AI system, trained on the life stories of millions of people, has showcased remarkable accuracy in predicting individual life outcomes and even the ominous time of death. While its efficacy in forecasting various aspects of life is undeniable, ethical concerns persist, making its use by life insurance companies a contentious issue. Moving forward, further research will help uncover the full potential and implications of this groundbreaking AI system.
