Scientists Discover that Artificial Intelligence can Predict Real Events in People's Lives

Artificial intelligences developed to model written languages can be used to predict events in people's lives. This was the discovery of a recent research project that involved several European and American universities.

The research has shown that if you use large amounts of data about people's lives and train the so-called "transformative models" (similar to the famous ChatGPT), which are used to process language, artificial intelligence is able to organize data systematically and predict what will happen in a person's life - even giving estimates of when they may die.

To reach this conclusion, the researchers analyzed health and labor market data from 6 million Danes using an artificial intelligence model called Life2vec. After the model was trained at an early stage and learned the patterns of the data, the researchers demonstrated that the model surpasses other advanced neural networks and can predict results such as personality and time of death with high accuracy.

Life2vec predictions work as answers to general questions, such as "What is the chance of death within four years?". When the researchers analyzed the responses of the model, the results were consistent with what is known through other research - For example, individuals who occupy a leading position or with high incomes are more likely to survive, while people with a mental diagnosis are associated with a higher risk of death.

The fact is that human life is like a long sequence of events - something similar to the functioning of a sentence in a language, which is a long series of words. And it is exactly for this type of task that the transformative models were created, which explains their high effectiveness

The ethics of artificial intelligences using data from real people

The researchers behind the article point out that there are still many ethical issues around the use of the Life2vec model, such as the protection of confidential data, privacy and the role of prejudice in data. These issues should be taken into account before the model can be used openly, for example, to assess the risk of an individual contracting a disease or any other life event.

Similar technologies are already being used within technology companies that, for example, track people's behavior on social networks, profile and use this information to display extremely well-targeted ads. According to scientists, a democratic conversation is needed so that we can understand where this technology is taking us, and if this is something we want.

The next steps of the research involve incorporating other types of information into the artificial intelligence model, such as texts, images or information about our social connections. This use of data will open the door to a new interaction between the social sciences and the health area.

News reference: Germans Savcisens, Tina Eliassi-Rad, Lars Kai Hansen, Laust Hvas Mortensen, Lau Lilleholt, Anna Rogers, Ingo Zettler, Sune Lehmann.Using sequences of life-events to predict human lives. Nature Computational Science, 2023.