What I think I will remember most about this interview a year from now is the discussion regarding ethics in genetics-based research. That researchers may have to choose whether or not to inform research participants when the data predicts an unfavorable outcome was not a dilemma I had previously considered, and I was surprised to hear from De Vivo that labs may “take coverage” from such responsibility since they are collecting data and are not themselves real doctors. However, I was relieved to hear that more recent research is more upfront in asking participants if they’d prefer to be informed in such cases.
Regarding an aspect of the interview that may affect society's future, I am curious about how these predictive models are communicated to the general public. Though researchers are aware of genetic factors which predispose individuals to certain ailments, that knowledge does not necessarily translate to methods which can allow said individuals to lessen their chances of falling victim. Yet much of the general public may believe otherwise, and companies prey on the public’s desire to know of these genetic factors without informing them that knowledge will not allow them to increase their chances of survival, taking data from researchers in order to create and sell pseudo-diagnostic tests.
https://www.labxchange.org/library/items/lb:HarvardX:15f6a2e5:lx_simulation:1
I agree that ethics are a super important part of the future of prediction. In some sense, the mathematics and logic behind making a good prediction is the easier part of a prediction, because the moral and ethical implications are much harder to objectively reflect upon. That is, we can easily create a loss function to measure our prediction's accuracy, but it's hard to quantify (and maximize) how "good" an action is in a moral sense.