This is a matter of life and death – no doubt about it. bot it.
Most people aren’t in a huge hurry to know when the big bite is going to bite, but a newly developed AI death calculator can now predict when a person will die with eerily accurate accuracy.
“We use the technology behind ChatGPT (what we call the Transformer Model) to analyze human life by representing each person as a series of events that occur in their life,” December 2023. said Sune Lehmann, lead author of the study.Predicting human life using a sequence of life events” he told the Post.
In their report, the professor of networks and complex systems at the Technical University of Denmark and his co-authors describe a method known as “life2vec,” which uses selected details of an individual’s life, such as income, occupation, place of residence, and health history. Introducing the algorithm that can be used. Determines life expectancy with 78% accuracy.
“We take advantage of the fact that human life in some ways shares similarities with language,” Lehman explained. “Just as words follow one another in a sentence, events follow one another in human life.”
It’s a little different than ChatGPT. It’s the ever-popular bot that tech wizards are employing to land their dream job or curate the perfect outfit. By closely examining a man or woman’s past, life2vec can calculate the outcome of that person’s life.
“This model can predict almost anything,” Lehman told the Post. He said his research team also uses this specialized program to predict people’s personalities and international movement decisions.
“We predicted death because that’s what people (insurance companies, for example) have been doing for years,” he added. “So we were very aware of what could happen.”
From 2008 to 2020, Lehman’s team studied a heterogeneous population of 6 million Danes of different genders and ages. Analysts used life2vec to identify participants who were likely to survive at least four years after January 1, 2016.
“The scale of our dataset allows us to represent individual human life trajectories at the sequence level, providing a detailed representation of how each person moves over time,” the report said. is written. “We’re looking at how an individual’s life evolves across different types of events (information about heart attacks mixed with information about salary increases and moving from urban to rural areas). You can observe it.”
Researchers used plain language such as “In September 2012, Francisco received 20,000 Danish kroner as a guard at Elsinore Castle” and “Hermione followed in her third year at secondary boarding school.” was used to enter AI-specific information about each study participant. 5 elective classes. ”
We then assigned different digital tokens to each piece of data, all of which were categorized very specifically. For example, a forearm fracture is represented as S52. Working in a tobacco store is coded as IND4726 and income is represented by 100 different digital tokens. And “postpartum hemorrhage” is O72.
life2vec used the information provided to predict who would die by 2020 almost perfectly over three-quarters of the time.
Research shows that factors that can contribute to early death include being male, having a mental health diagnosis, and working in a skilled occupation. Earning a higher income or holding a leadership role were both associated with longevity.
However, Lehman stressed to the Post that study participants were not given a prediction of death.
“That would be extremely irresponsible,” he said, adding that he and his team ultimately hope to share details of the results in a way that protects the privacy of study participants.
“But we can still learn from it [life2vec] What are the factors that might help people live longer?” Lehman said. “We haven’t delved too deeply into this, but this is another important application of the model.”
Currently, this bot is not available to the general public or businesses. And even if it were to be deployed at scale, this AI would not be used to notify specific individuals in cases such as writing insurance policies or making hiring decisions. Probably not, says Prober.
“Forecasting is not used for anything,” Lehman argued. “The point of life2vec is to understand what is predictable and what is not.”
Source: nypost.com
