Predicting Death with Artificial Intelligence
The use of artificial intelligence (“AI”) is saturating all facets of life and death. While we might often think of AI as some future product of a technologically advanced society, it is already in common use. Think of Apple’s Siri and Google Translate; both require AI in order to function.
Earlier this year, my colleague, Garrett Horrocks, blogged on a study showing the promising use of AI in detecting Alzheimer’s. This month, a recent study out from the University of Nottingham explores the use of AI in predicting premature death of middle-aged persons. The study shows promising results.
AI and Bias
While many reports are optimistic in how such predictive models can improve preventative health care, others are more cautious. A recent article from Wired raises the issue of potential bias in such AI models. The article delves into the concerns of scholars that AI might adopt and even promote bias as a result of implicit biases that already exist. Take, for example, the Amazon AI recruitment tool which was designed to review resumes of job applicants and pick the top candidates. Amazon abandoned the project after experiencing several issues, including the program explicitly discriminating against women. The program did so by penalizing candidates who graduated from women’s colleges or had the word “women’s” in their resume (e.g. “women’s chess club”).
The Wired article also raises concerns about existing biases in health care services, such as how patients of different ethnics groups are treated differently for pain with studies in the US finding that racial and ethnic minorities tend to be undertreated for pain, compared to non-Hispanic white persons. While the Wired article raises concerns about the potential biases that can be adopted and/or promoted by AI, the article also notes the potential for AI to reduce bias by focusing on objective factors affecting a person’s health.
AI and the Law
Many say that the law and lawyers are resistant to change (who still relies on faxes?). Despite any such resistance, the legal system, like everyone else, is being dragged into the world of AI, whether ready or not. Just as AI is revolutionizing health care, legal products implementing AI are being developed, with some estimating that over 100,000 jobs in the legal sector will be automated by 2036.
More importantly, however, is the ongoing need for the law to adapt to the changing world of AI. The implementation of AI in our everyday life has significant ramifications from the products recommended to us while online shopping to whether or not we might receive proper preventative health care. With the potential for ethical abuses and unintended consequences (such as discrimination), it will be interesting to see how (or if) laws and regulations develop to address these new advances in AI.
Thanks for reading!