No Safe Zone
Insurance is a long standing citadel in the United States. The companies that offer it are paragons of the financials industry offering funds to people when disaster strikes. However, this is not the case anymore. WMDs are prevalent in the insurance industry; they take proxy data and use it to stratify the population into groups that are somewhat like each other, and determine their risks from their. Higher risk means higher premiums. Unfortunately, these systems fall into the category of WMDs because they utilize seemingly irrelevant data, like a person’s credit score, to quantify their risk of a car accident or a visit to the emergency room.
To combat these generalizations, there is some things people can do. In terms of health insurance, offering to give in to wellness programs that track health data in real time pays dividends in terms of reduced premiums. In terms of car insurance, offering to give in to “Safe Driver” programs that track car telemetry and GPS data in real time reduces premiums; however, these are just the start to these problems. As these companies amass more data, more specific categories can be built, and in the case of All State, a large insurance company in the United States, that is already happening.
The issue with these mathematical models is that they heavily rely on proxies to make assumptions about people. A bad credit score does not equal a poor driver, a low GPA in high school does not equal an unhealthy individual, but to a mathematical model, these sorts of correlations crop up. An interesting study brought up by O’Neil, was the case of the company Sense. They developed a “blind” mathematical model that was fed data, but without much insight as to what it really means. By having no weighting or biases implemented by the developers, the model returned curious results that grouped people into tribes on criteria that humans would never expect. These types of models are superior to their WMD counterparts and do not prey on the poor and ethnic minorities.
This chapter in O’Neil’s book expanded on her idea that WMDs are ubiquitous. There is little in life that someone can do without being graded and generalized by a WMD into some social subcategory labelled “desirable” or “undesirable.” There is a need to enact change in these sectors. Establish rules and regulations that assure that models that predict risks via inputs that are proxies for geographical location, social status, and skin color are illegal, as they are when done by hand.
