Artificial intelligence (AI) is not just limited to writing articles, business proposals and even movie scripts. It could even predict a person’s life span. Researchers have developed an AI tool that uses life events like health history, education, occupation and income to predict an individual’s personality to lifespan, PTI reported.
The AI tool called life2vec has been designed using transformer models which power large language models like OpenAI’s ChatGPT. This tool is trained on a data set pulled from the entire population of Denmark.
According to researchers, life2vec can predict future including the lifespan of individuals with an accuracy exceeding state-of-art models. But the research team said it is best used as the foundation for future project, not an end.
ALSO READ: Sam Altman-backed Humane to launch wearable AI Pin: Check features, prices, release date
Tina Eliassi-Rad, a professor at United States’ Northeastern University, said that the researchers are using prediction to evaluate how good such AI models are, but life2vec should not be used for prediction on real individuals.
Rad added that the AI model is based on a specific data set of a specific population. Social scientists have been involved in the process of developing this AI tool, and the team is hoping that it will bring a human-centric approach to AI development.
Sune Lehmann, author of the study published in the journal Nature Computational Science, was quoted by PTI as saying,”This model offers a much more comprehensive reflection of the world as it is lived by human beings than many other models.”
ALSO READ: One Piece creator Eiichiro Oda stirs controversy with AI-generated song
The researchers used data set to create long patterns of recurring life events to feed into life2vec. A transformer model approach was used to train the LLM on the language and adapting it for a human life represented as a sequence of events.
“The whole story of a human life, in a way, can also be thought of as a giant long sentence of the many things that can happen to a person,” Lehmann added.
By using information learnt from observing millions of life event sequences, life2vec built vector representations in embedding spaces. It then begins to categorise and draw connections between life events like income, education, or health factors.
These embedding spaces serve as a foundation for the predictions the model ends up making. One of the life events that the researchers predicted was a person’s probability of mortality, the report added.
Lehmann said that the when the team visualises the space used by the AI model to make predictions, it resembles a long cylinder that takes you from low probability of death to high probability of death.
