New AI Model Predicts Potential Lifespan And It Is Eerily Accurate

While most people are not in a rush to find out when they will pass away, a group of researchers have developed an artificial intelligence (AI) tool that can predict an individual's lifespan based on income, job, education, health history and other life events.

The AI tool draws its power from transformer models, which are used by ChatGPT-like large language models (LLMs) to run. Also, it is trained on a data set derived from the entire population of Denmark.

"We use the technology behind ChatGPT (something called transformer models) to analyze human lives by representing each person as the sequence of events that happens in their life," Sune Lehmann, lead author of the December 2023 study told The New York Post.

The tool uses this sequence of life events to predict human lives. In the report, the professor of network and complex systems from the Technical University of Denmark and co-authors unveiled an algorithm called Life2vec, which is capable of determining an individual's life expectancy with 78 per cent accuracy.

According to Lehmann, human lives share a similarity with language in a certain sense. Shedding some light on his statement, he said: "Just like words follow each other in sentences, events follow each other in human lives."

"This model can predict almost anything," Lehmann told The Post. The researchers decided to predict death because it is something people have worked on for a long time (for instance, insurance companies), Lehmann explained.

Reportedly, Google is working on a similar program codenamed Project Ellmann, which will give users a "bird's-eye" view of their lives.

How reliable is Life2vec?

Despite its predictive power, researchers recommend using Life2vec only as a foundation tool for upcoming work rather than an end in itself. According to Tina Eliassi-Rad, who is a professor at Northeastern University, US, the tool shouldn't be used for predictions on real people.

She points out that the tool is just a prediction model that uses a specific data set of a specific population. However, the researchers are sparing no effort in a bid to bring a human-centred approach to AI development.

In line with this, they included social scientists in the process of building Life2vec. "This model offers a much more comprehensive reflection of the world as it is lived by human beings than many other models," Lehmann said.

Lehmann's team reportedly examined a heterogeneous subject population of 6 million Danish people between 2008 and 2020. It is worth noting that these people varied in sex and age.

With the help of Life2vec, the analysts tried to figure out which of the subjects are likely to live for at least 4 years beyond January 1, 2016.

"The scale of our dataset allows us to construct sequence-level representations of individual human life trajectories, which detail how each person moves through time," the report noted.

This is a major step forward towards using AI to improve human life. In fact, a new study found that a considerable number of Americans are relying on AI to improve healthcare in 2024.