A recent study brought to light the remarkable abilities of an AI model akin to ChatGPT, trained on the life stories of over a million individuals, accurately predicting both the trajectories of individual lives and the likelihood of early mortality.
Scientists at Denmark Technical University (DTU) harnessed personal data from over a million people in Denmark to train this AI model. Remarkably, it surpassed existing systems in predicting individuals’ chances of death with exceptional precision.
By analyzing data encompassing education, medical visits, diagnoses, income, and occupation of six million individuals from 2008 to 2020, collected from health and labor markets, researchers converted this information into textual data to train a sophisticated language model called “life2vec,” akin to the technology used in ChatGPT-like applications.
Published in the journal Nature Computational Science, the study highlighted how this AI model outperformed other systems by learning intricate patterns within the data, accurately forecasting outcomes such as personality traits and estimated time of death.
In one instance, researchers tasked the AI with predicting survival outcomes for a group aged between 35 and 65, half of whom had passed away between 2016 and 2020.
Astonishingly, the AI’s predictions were 11% more accurate than those of other AI models or the methods employed by life insurance companies to set policy prices.
Lead author Sune Lehman from DTU explained, “We aimed to explore the extent to which we can foresee future events based on past conditions and events in individuals’ lives.”
He emphasized that the model’s precision lay in specific data aspects, rather than merely the existing predictive frameworks.
The researchers also noted that their framework opens doors for identifying new factors influencing life outcomes and potential avenues for personalized interventions, thus signifying a significant advancement in predictive modeling for individual life trajectories.
I am an experienced writer, analyst, and author. My exposure in English journalism spans more than 28 years. In the past, I have been working with daily The Muslim (Lahore Bureau), daily Business Recorder (Lahore/Islamabad Bureaus), Daily Times, Islamabad, daily The Nation (Lahore and Karachi). With daily The Nation, I have served as Resident Editor, Karachi. Since 2009, I have been working as a Freelance Writer/Editor for American organizations.
