While most people are hesitant to find out when they will “go to heaven”, a new artificial intelligence system using ChatGPT technology promises to predict the time of death of a person with great accuracy.
According to a study published in the journal Nature Computational Science, the pioneering artificial intelligence system “life2vec”, trained on the life histories of more than a million people, highly accurately predicts life expectancy as well as the risk of premature death with 78% accuracy.
The AI model was trained on the personal data of the Danish population and showed that it predicts the likelihood of people dying more accurately than any existing system, reported scientists from the Technical University of Denmark (DTU).
During the study, scientists analyzed data on the health and type of work of 6 million Danes, collected from 2008 to 2020, including information on people’s education, doctor and hospital visits, diagnoses, income and occupation. Scientists converted the dataset into words to train a large language model called life2veca similar technology behind artificial intelligence applications such as ChatGPT.
“We use the technology behind ChatGPT to analyze human life, representing each person as a sequence of events that happen in their life,” Sune Lehmann, lead author of the study, told the New York Post.
The researchers took data from a group of people aged 35 to 65, half of whom died between 2016 and 2020, and asked an AI system to predict who would live and who would die. They discovered that she predictions were 11% more accurate than predictions from any other existing AI model or methodused by life insurance companies to price policies.
Using the model, the researchers looked for answers to general questions, such as the likelihood of a person dying within 4 years. They found that the model’s responses were consistent with existing findings, for example that when all other factors are taken into account, people in management positions or with high income are more likely to survive, while being male or having a mental health diagnosis is associated with a higher risk of death.
“We used this model to answer the fundamental question: to what extent can we predict the events of your future based on the conditions and events of your past, Lehmann said. “From a scientific point of view, we are not so much interested in the forecast itself, but rather in the aspects of the data that allow the model to give such accurate answers.”
The model can also accurately predict personality test scores in a specific segment of the population better than existing artificial intelligence systems. “Our the system allows researchers to identify new potential mechanisms affecting lifeand the associated potential for personalized interventions,” the researchers wrote.
Treating each part of your life like words in a sentence, life2vec predicts how the story will end based on what has been written so far.
Just as ChatGPT users ask it to write a song, poem or essay, scientists can ask life2vec simple questions such as “how likely is death within four years?” for a specific person. Based on population data, he correctly predicted who would die by 2020 more than 3/4 of the time.
Just as ChatGPT and other large language models were trained on a variety of existing written works, life2vec was trained on data from people’s lives, written as a series of sentences. These include sentences such as “In September 2012, Francisco received twenty thousand Danish kroner as a guard at Elsinore Castle” or “In her third year of high school, Hermione took five electives.” Lehmann and his team assigned different points to each piece of information, and all this data was compared with each other.
Categories in people’s life histories span the entire spectrum of human experience: a broken forearm or something like that. Occupation is scored: for example, working in a tobacco shop is coded as IND4726, income is coded as 100 different numbers, and “bleeding during childbirth” is coded as O72.
Many of these relationships are intuitive, such as certain activities bringing in more money, providing better nutrition and better health/early disease detection. And working in hazardous enterprises shortens life. But what life2vec does is analyze the vast mosaic of factors that make up a person’s life. And in the end it makes a forecast based on millions of data from other people.
Artificial intelligence can also make predictions about a person’s personality. To do this, Lehman and his team trained a model to predict people’s responses to personality test questions. The test asks respondents to rate 10 items based on how much they agree, such as: “The first thing I always do when I get to a new place is make friends,” or “I rarely express my opinions in group meetings.”
However, scientists warn that this model should not be used by life insurance companies for ethical reasons.
More Stories
Where will Greeks go for Easter: the best domestic and foreign destinations
Fine 1200 euros for the bad habit of Greeks at Easter, who is at risk
UK: human trial of melanoma vaccine started