The “Paycheck”-style tool is highly accurate based on analyses of people's personal data.

A team of researchers from four universities on three continents has developed an artificial intelligence tool that - based on analyses of people’s personal data - can predict events in their lives and even the time of their death.

Acting much like ChatGTP, the “death calculator” is a transformer model utilizing natural language processing techniques. To learn the art of future forecast, the tool - called Life2vec - was fed data from 6 million individuals who live in Denmark, including daily information about their health, education, occupation, income, address, and working hours.

More to read:
Can humans defeat death? Science says “maybe”

The team, which employed scientists from the Technical University of Denmark, the University of Copenhagen, the Istanbul Technical University, and the Northeastern University in Boston said in a press release that they examined how human lives evolved for some time and analyzed the detailed sequences of events to build up anticipations. 

The researchers organized life events into a stable and structured "space" or system.

The model's predictions align with existing social science findings; for instance, individuals in leadership positions or with high incomes are more likely to survive, while being male, skilled in a specific area, or having a mental diagnosis is associated with a higher risk of mortality.

More to read:
[video] Australian doctor promotes euthanasia pod for voluntary suicide

The models can predict various outcomes, from early mortality to personality traits, outperforming existing models significantly - similarly to the plot of the 2003 science fiction movie “Paycheck” with Ben Affleck and Aaron Eckhart. 

The researchers see human life as a sequence of events, akin to a sentence in language, and apply transformer models to analyze these life sequences, they said in a study published in the journal Nature Computational Science. 

While the study raises ethical and privacy concerns about knowing the time of one's death, the researchers suggest the next steps involve incorporating additional information, such as text, images, or details about social connections.

More to read:
Billionaire Peter Thiel signs up for cryonic program after his death

Playing the oracle isn’t happening for the first time. In 2018, Stanford University successfully tested an AI tool that estimating the time a patient may die, claiming that it was designed to help physicians prompt patients to have necessary end-of-life conversations with their families. 

A year later, University of Nottingham published a similar research on AI-assisted premature death forecasting.