EECS 649 Blog 11: No Safe Zone

Author: Megan Rajagopal

In Weapons of Math Destruction, O’Neil talks about big data and how models impact people and their ability to get insurance. O’Neil mentions how companies will use big data to determine the risk of future clients. In this chapter, O’Neil looked at how big data is used in relation to insurance companies. Insurance companies will collect large amounts of data for every person they are working with. Her specific example looks at how big data is used to detect risk for future clients who are looking at purchasing car insurance. They collect information like location, commute length, how well they drive, etc. They will use all of this data and generate a model that will create their own e-score that will determine the risk of a client, the risk in this case would be how responsible of a driver they are. However, these e-scores have demographic data weighing higher than a client’s driving record; which is very interesting considering the models is for car insurance. It was also noted that the client’s credit scores also were considered. An example of credit scores weighing more than driving record is mention in Florida, adults with a clean driving record and poor credit scores paid around $1552 more than drivers with excellent credit scores and a drunk driving conviction. Furthermore, insurance companies will also use the big data they collected and group similar clients together. For example, they will look at all the clients who live in a certain zip code and use that information to generate the risk of a future client. However, since they are using data that clients may not be able to control (such as demographic data) there is a chance that clients are not going to be accurately represented. With that said, this creates a negative feedback loop because marginalized people or people who live in poorer communities will continue to be negatively impacted, meaning these models are weapons of math destruction. One way that the car insurance system can be fixed is making sure car insurance companies only look at driving records to determine the risk of a client. Using driving infractions instead of location, credit sores, race, gender, etc. will allow every client to be judged fairly. In general this chapter was very insightful and I remember experiencing the weapons of math destruction with car insurance when I was a young driver. My parents told me that students who got good grades had cheaper car insurance. I was also told that teenage girls were more likely to have cheaper insurance than teenage boys, because data and science suggested that girls were less reckless than boys. Overall, it is important to see how all your data is being used by insurance companies. Companies will continue to take data from clients to improve their models but the models may not be fair to everyone. All in all, it was beneficial to learn how my data is used in the insurance world so I can try my best to fix and improve the data I can control so I can try and prevent getting exploited by the unfair models.