Researchers claim this AI knows when you will die

Researchers have developed an AI system they claim can predict the mortality of humans more accurately than current state-of-the-art models.

The AI model was trained on the data of millions of Danish citizens.

(Matthew Modoono/Northeastern University)

Scientists have built an artificial intelligence tool that uses people’s details — like health history, education, job and income — to predict everything from their personality to their mortality. The new tool is built using transformer models similar to those that power large language model-based tools like ChatGPT.

The AI tool, called life2vec, was trained on a data set pulled from the entire population of Denmark, which was made available only to the researchers by the government of the country. According to Northeastern University, the tool built using complex datasets can predict the future, including lifespans, with an accuracy that is higher than state-of-the-art models.

“Even though we’re using prediction to evaluate how good these models are, the tool shouldn’t be used for prediction on real people. It is a prediction model based on a specific data set of a specific population.” said Tina Eliassi-Rad, professor of computer science at Northeastern University, in a press statement.

Story continues below this ad

The researchers involved AI ethics experts and other social scientists on to the project, hoping to bring a human-centric approach to AI development. “These tools allow you to see into your society in a different way: the policies you have, the rules and regulations you have. You can think of it as a scan of what is happening on the ground,” added Eliassi-Rad.

Also read | Scientists simulate asteroid defence using nukes

As you can expect, the massive data set used to train life2vec is at the heart of the study and it spells out the many events and elements that make up the life of Danish citizens, including health factors and education to income. The researchers used the data to create patterns of recurring life events to feed into the model.

According to Sune Lehmann, co-author of the study published in the journal Nature Computational Science this month, the whole story of a human life can be thought of as a giant long sentence of many things that happen to a person.

The model observes from millions of life event sequences in the training data and then builds what are called vector representations in embedding spaces, where it categorises and draws connections between life events like income, education or health factors. These embedding spaces are the foundation for the prediction the models make.

Story continues below this ad

“When we visualize the space that the model uses to make predictions, it looks like a long cylinder that takes you from low probability of death to high probability of death. Then we can show that in the end where there’s high probability of death, a lot of those people actually died, and in the end where there’s low probability of dying, the causes of death are something that we couldn’t predict, like car accidents,” said Lehmann in a press statement.

The study also showed that the model is capable of predicting individual answers to a standard personality questionnaire, especially when it comes to the quality of extroversion.

Tags:
artificial intelligence