Researchers in Denmark are utilizing artificial intelligence and data from a vast number of individuals to predict various stages of a person’s life, aiming to highlight both the capabilities and risks associated with this technology.
The creators of life2vec are not driven by morbid curiosity but rather seek to investigate the patterns and connections that deep-learning programs can unveil to forecast a broad spectrum of health and social milestones.
Sune Lehmann, a professor at the Technical University of Denmark (DTU) and a co-author of a study recently published in the journal Nature Computational Science, described life2vec as a versatile framework for predicting diverse aspects of human lives based on available training data.
The algorithm, inspired by ChatGPT, scrutinizes factors influencing life trajectories such as birth, education, social welfare, and employment schedules.
Drawing parallels to language-processing algorithms, the team endeavors to analyze the predictability and evolution of human lives through detailed event sequences.
While some have labeled the program as a “death calculator,” the researchers clarify that the software is currently private and not accessible online or to the broader research community.
The foundation of the life2vec model rests on anonymized data from approximately six million Danish individuals gathered by Statistics Denmark. By scrutinizing event sequences, the algorithm can forecast life outcomes up to the final moments of an individual’s life.
In predicting mortality, the algorithm boasts an accuracy rate of 78%, while it achieves a 73% accuracy rate in forecasting relocations.
The researchers primarily focus on predicting premature deaths within a specific age group (35-65 years) to validate the algorithm’s efficacy.
Despite its promising performance, the tool remains in the research phase and is not yet deployable outside experimental settings.
In contrast to the proprietary algorithms developed by tech giants, the researchers emphasize the importance of transparency and public engagement in exploring the potential implications of such data-driven models.
Pernille Tranberg, a Danish data ethics expert, underscores the significance of public awareness, particularly in light of businesses like insurance companies employing similar algorithms for decision-making, which could potentially lead to discriminatory practices.
While commercial algorithms claiming to predict life expectancy already exist, their reliability remains questionable, underscoring the need for caution in interpreting and utilizing such tools.
This research project serves as a scientific response to the growing influence of AI algorithms developed by major tech companies, offering a public platform to examine the possibilities and limitations of data-driven predictions.
