Is it possible that artificial intelligence can predict the most important facts of our lives? Studies, marriage, children and even our death, as if they were part of a linear sequence? Obviously not, one might reply, yet a research project conducted by the University of Copenhagen and Northeastern University in the United States would demonstrate the exact opposite. That is, if you use large quantities of data on people's lives and train so-called “transformation models”, which – like the famous ChatGPT, are used to process language – these can systematically organize the data, therefore predicting what will happen in a person's life. And even more distressing, even estimating the time of our death.
The scientific project, eloquently titled, 'Using Sequences of Life-events to Predict Human Lives', published in Nature Computational Science, is based on labor market data and data from the National Patient Registry and the Danish Statistical Office. The database includes all 6 million Danes over a time period from 2008 to 2020 and contains information on income, salary, type of job, industry, social benefits, while the healthcare dataset includes records of visits to healthcare providers or hospitals , diagnosis, type of patient and degree of urgency. Danish and American researchers analyzed all this data to train an artificial intelligence model, called life2vec.
“We used the model to answer the fundamental question: to what extent can we predict events in your future based on conditions and events in your past? Scientifically, what is interesting to us is not so much the prediction itself, but the aspects of the data that allow the model to provide such precise answers,” says Sune Lehmann, professor at the Danish University and first author of the paper.
One of the questions asked of the Life2vec model is: “Death within four years”? And the model provided answers consistent with what is already known in the field of social sciences at a statistical level. To understand better. If we take as an example, individuals in a leadership position or with a high income, they are statistically more likely to survive, while being male, qualified or having a diagnosis of a mental health problem is associated with a higher risk of death.
Life2vec encodes data into a large system of vectors, where the model decides where to place data related to time of birth, education, salary, housing, and health. “What's exciting is considering human life as a long sequence of events, similar to how a sentence in a language is made up of a series of words. This is usually the kind of task for which life transformation models are used 'artificial intelligence, but in our experiments we use them to analyze what we call life sequences, that is, events that happened in human life,' says Sune Lehmann.
According to the researchers, the next step is to incorporate other types of information, such as texts and images or information about our social connections, but this new knowledge needs to be understood more deeply, before the model can be used, for example, to evaluate an individual's risk of contracting a disease or other life events.
#Artificial #Intelligence #predict #lives