Researchers claim that a novel AI model has the potential to forecast the future, unveil human lifespan, and personality traits.

A team of researchers has created an artificial intelligence (AI) tool capable of utilizing a series of life events, including health records, education, employment, and income, to forecast various aspects of an individual’s life, ranging from their personality traits to their life expectancy.

A tool called Life2vec is built using transformer models that use large-scale language models (LLMs) such as ChatGPT.

Life2vec can predict the future, including the lifespan of individuals, with an accuracy that exceeds state-of-the-art models, the researchers said. Despite its predictability, the research team said it is best used as a basis for future work, not as an end in itself.

“Although we use prediction to evaluate how good these models are, the tool should not be used to predict real people,” says Tina Eliassi-Rad, a professor at Northeastern University, USA.

“It’s a predictive model based on a specific set of data from a specific population,” Eliassi-Rad said.

By involving social scientists in building this tool, the team hopes it will bring a human-centered approach to AI development that doesn’t lose sight of humans amid the vast datasets their tool is trained on.

“This model provides a much more comprehensive reflection of the world as people live in it than many other models,” said Sune Lehmann, author of the study published in the journal Nature Computational Science.

At the heart of Life2vec is a huge dataset that researchers used to train their model.

The researchers used this data to create long models of repeated life events that they feed into their model, using the transformer model approach used to train LLMs in the language and adapting it to human life, represented as a series of events.

“The whole story of human life can also be thought of as a giant long sentence about the many things that can happen to a person,” said Lehmann, a professor at the Technical University of Denmark.

The model uses the information it learns from observing sequences of millions of life events to build so-called vector representations in embedded spaces, where it begins to classify and draw connections between life events such as income, education or health factors.

These embedding states serve as the basis for the predictions the model ultimately makes, the researchers said.

One of the life events predicted by the researchers was a person’s probability of dying.

“When we visualize the space that the model uses to predict, it looks like a long cylinder that takes you from a low probability of death to a high probability of death,” Lehmann said.

“Then we can show that in the end, where there’s a high probability of death, many of them actually died, and in the end, where there’s a low probability of death, the causes of death are something we couldn’t predict, like car accidents,” the researcher added.