A Humbler Data Science

As in all quandaries concerning privacy, accuracy, individuals rights, and technology it is not feasible to approach this question in the view of black and white. As such, my response possesses many qualifiers as I approach this article. Generally, I support the widespread use of data science in the future. I believe there are many sectors of the world and society that would improve if it was applied correctly. For instance, the potential humanitarian improvements outlined in the article would be incredibly impactful if implemented in a responsible manner. The first example presented in the article seems fairly uncontroversiaL: using satellite images to track destruction from natural disasters in order to coordinate relief efforts. This is clearly adding benefit to society and given the images released are for a very specific area at a specific time, the possibility of misuse if low. However, even in instances in which people or organizations are attempting to improve a section of society, there are unforeseen consequences. For instance, the article discusses the fact that some organizations use satellite data to determine which people have thatched roofs, as this has been found to correlate to being poor. Then the organizations can dispatch relief in these areas. At its core, this is a morally good endeavor, but people have nonetheless found a way to abuse it. In some areas, families pretend to live in a building with a thatched roof that is next to their larger, actual home without a thatched roof. In this way, people who do not require relief can take it away from those who desperately need it. This is an example of the way in which data science can fail because of an outside entity; it can also fail due to the scientists. Data science is inherently inexact as absolutes do not exist in any science. Everything is either a prediction or hypothesis, and even a well-tested hypothesis is subject to change as new data comes to light. As such, the conclusions drawn and algorithms made cannot be considered fact for all time. The article illustrates the capricious nature of data science when mentioning that an algorithm that is effective at predicting rates of poverty or affluence in an area could become ineffective only a month later. Evidently, the world cannot move into a place where we rely solely on data to assess situations and predict outcomes or the world will be in danger of wrong assertions very frequently. Nevertheless, I believe the use of data science in humanitarian endeavors is very important because ultimately the good produced outweighs the potential bad. Organizations must take care to use data science as an auxiliary tool and not as the sole approach to solving a problem and must implement this is an ethically responsible way in order to maintain this balance.

There are uses of data science that I am squarely opposed to, such a the use of “social credit” scores in China. I am not that obsessed with privacy and acknowledge the fact that the government and private companies know everything about me, and I am even able to see the upside of some of that. But I see the outright discrimination of people based on data collected as a severe breach of human rights (yes, the US government definitely does this). Additionally, clearly the misuse of data by authoritarian regimes is not morally righteous or beneficial. Therefore, although what is good and what is bad is subjective, the general rule is to work in order to better society, but in a responsible manner.