Knowing when you’re going to die is not a comforting thought to many. However, gaining some perspective on the finite amount of time you’ve got to enjoy all the things and people you love has been shown to have positive mental health effects.
So it’s with some caution that you might want to try the new research project from the Technical University of Denmark (DTU) which claims to have invented a tool able to predict an individual’s major upcoming life events, and even their death.
The project is a collaboration between DTU, the University of Copenhagen, ITU, and Northeastern Univerisity in the US. They have shown that if you input large amounts of statistical health and life-outcome data into large language model (LLM) style artificial intelligence programmes, they can organise and analyse the data reliably to predict future events.
Of course, if you wanted to find your own results, you’d have to be Danish. The teams input health and labour market data for 6 million Danes into the model they’ve called Life2vec, including industry information, social benefits, GP visits and other information across a span of 12 years. The results are thought to be more accurate and advanced than other neural networks currently available, according to their study published in Nature Computational Science.
“We used the model to address the fundamental question: to what extent can we predict events in your future based on conditions and events in your past?,” Professor Sune Lehmann has said.
“Scientifically, what is exciting for us is not so much the prediction itself, but the aspects of data that enable the model to provide such precise answers.”
Can AI Predict Your Death?
The predictions of Life2vec deal with general questions like ‘who will die within four years?’ They state that the results align with what other purely data-based predictions indicated by health and labour information would suggest.
The researchers state that their AI model is more a proof of concept than a specific tool for prediction by interpreting human life in the same way AI would interpret language patterns. The fact that it appears to work at all opens up a “whole new interaction between health and social sciences.”
“What’s exciting is to consider human life as a long sequence of events, similar to how a sentence in a language consists of a series of words,” Lehmann said.
“This is usually the type of task for which transformer models in AI are used, but in our experiments we use them to analyze what we call life sequences, i.e., events that have happened in human life,” says Sune Lehmann.
Pretty much since the dawn of AI technology, people have hoped that it can be used to predict future events. Using a programme to take the guesswork out of finding your ideal romantic partner, for example, is something that companies already claim to be able to do. Others are trying to use it to predict stock market moves and make bank.
Predicting impending mortality, however, comes with its own set of ethical issues. Previous attempts to predict death in a population have been made with some success. In one US study, an AI programme was able to predict the likelihood of death using electrocardiogram data — with accuracy that even its inventors couldn’t explain. Exactly what you do with that information is not always clear.
Of course, AI is not going to be able to predict if you’re going to be hit by a bus tomorrow. Instead, it uses granular historical information to determine how long someone with the same health and life history lived. That’s something that many people want to know in the hopes that they can change their lifestyle factors to push that date back as far as possible. It’s why the fitness tracker market is worth $67 billion globally.
The Danish team say that their next steps will be to adapt the model to be able to interpret visual, audio, and social information to make it even more accurate in the hopes it can be used to help people shift their behaviours. Because, as they point out, big tech companies are already harvesting this information on us, giving them a greater understanding of our own mortality than us.
“Similar technologies for predicting life events and human behaviour are already used today inside tech companies that, for example, track our behaviour on social networks, profile us extremely accurately, and use these profiles to predict our behaviour and influence us,” Lehmann said.
“The model opens up important positive and negative perspectives to discuss and address politically.
“This discussion needs to be part of the democratic conversation so that we consider where technology is taking us and whether this is a development we want.”
Related: In China, Mourners Are Resurrecting the Dead Using AI
Related: Online Dating’s About to Be Made a Whole Lot Better — or Worse — With AI
Read more stories from The Latch and subscribe to our email newsletter.