An algorithm trained on the data of millions of people can read the future with a precision that would be the envy of fortune tellers and fortune tellers.
The AI system uses the ability to recognize recurring patterns in large amounts of data to predict the evolution of a person’s life, from salary increases to the possibility of early death. The (somewhat disturbing) feats of the fortune-teller algorithm were described in the scientific journal Nature Computational Science.
Table of Contents
The study on the algorithm that predicts the future
Sune Lehmann Jørgensen, a network scientist at the Technical University of Denmark, started from the data contained in a rich Danish database – the Danish national registers, which includes information on the work and health of around 6 million citizens – and converted them into strings of text, which then fed Life2vec a large language model (Large Language Model, the same category as the ChatGPT chatbot).
These systems are capable of identifying, within large amounts of text, the recurring patterns in the succession of words and sentences, and of using them in turn to process coherent texts.
Scientists therefore wondered whether these same abilities could be exploited to trace a possible trajectory in other sequences – those of significant events in our lives. The first step was therefore to translate details such as salary size, job title or medical diagnoses into concise text strings, such as: “In August 2010, Agnes earned 30,000 Danish kroner as a midwife in a hospital in Copenhagen.”
The crystal ball algorithm
The model was trained on the lives of every individual who entered the database between 2008 and 2016, and this information was used to ask the algorithm to “predict” how many of the people studied would die by the year 2020 (the “future”, compared to the era on which the AI had trained).
The AI’s predictions proved accurate in 78% of cases. The system was able to identify various factors that favored a greater risk of premature death, such as having a low salary, having had a diagnosis of mental illness, being male, even if it “missed” some less predictable causes of early death such as heart attacks or accidents.
Risks or benefits?
The algorithm also correctly predicted other aspects of people’s lives, such as the propensity for certain character traits (easily associated with different types of jobs).
The model could be exploited for good, for example to imagine the future risk of developing certain diseases and act accordingly. But it could also pose major risks to the privacy, security and rights of citizens.
«Clearly it should not be used by an insurance company, because the basic idea of insurance is that by sharing the uncertainty about who will be the unfortunate individual affected by an accident, death or loss of a suitcase, you divide this load” comments Jørgensen.
Read also: Google boosts artificial intelligence with new smartphone serie
