Intriguing questions about the duration of our lives often occupy our thoughts, prompting a natural curiosity about our life expectancy. Capitalizing on the advancements in artificial intelligence (AI) technology, researchers have delved into unraveling these mysteries. A recently developed AI death calculator claims to predict an individual’s death with a purportedly chilling accuracy of 78 percent.
Spearheading this innovative approach is Sune Lehmann, the lead author of the study titled “Using a sequence of life-events to predict human lives,” published in December 2023. Lehmann, a professor specializing in network and complex systems at the Technical University of Denmark, along with his co-authors, introduces the algorithm “life2vec.” This AI model employs the underlying technology behind ChatGPT, known as transformer models, to analyze human lives by representing them as a sequence of life events.
The algorithm utilizes selective information from an individual’s life, incorporating factors such as health history, residence, income, and profession to estimate life expectancy, asserting an impressive 78 percent accuracy. Lehmann emphasizes the analogy between human lives and language, noting that just as words follow each other in sentences, events follow each other in human lives.
While life2vec differs from the well-known ChatGPT, a bot aiding individuals in job searches and fashion choices, it specializes in computing life outcomes by meticulously examining a person’s past. Lehmann underscores the model’s versatility, claiming it can predict a wide array of outcomes. The research team employs specialized programs to forecast personalities and, notably, death, drawing on decades of work in areas such as insurance.
The study delves into a heterogeneous subject population comprising approximately 6 million Danish individuals of various genders and ages from 2008 to 2020. Utilizing life2vec, analysts sought to identify subjects likely to survive for at least four years after January 1, 2016.
The extensive dataset enables the construction of sequence-level representations of individual human life trajectories. The report highlights the observation of how individual lives evolve in a space of diverse event types, where information about health incidents intersects with salary increases or changes in residence.
Researchers fed the AI-specific information using simple language examples, such as “In September 2012, Francisco received 20,000 Danish kroner as a guard at a castle in Elsinore” or “During her third year at secondary boarding school, Hermione followed five elective classes.”
Lehmann’s team’s groundbreaking work raises ethical considerations about the potential implications of predicting personal outcomes, especially the timing of one’s death. The convergence of AI and life predictions poses both opportunities and challenges, with potential applications extending beyond mortality forecasts.
As technology continues to advance, the intersection of AI and human life prompts intriguing questions about privacy, consent, and the responsible use of predictive technologies. While this AI death calculator underscores the capabilities of current technology, it also opens a dialogue about the ethical boundaries and societal impact of delving into the predictions of deeply personal and profound aspects of human existence.
