Online Advertising — Weapons of Math Destruction

Daniel Johnson
2 min readMar 14, 2021

I was tasked by my artificial intelligence class to write about chapter 4 in Cathy O’Neil’s book, Weapons of Math Destruction. This chapter is on nefarious practices in online advertising.

O’Neil starts out the chapter by describing the base level of how online, personalized, advertisements should work. She uses a good example of of a pizzeria targeting people when it is a sport game day. This is pretty harmless and is a good way to get a product to people who are more likely to want it. However, she quickly goes into many examples of much worse practices of these sorts of advertisements.

A big theme she keeps throughout this chapter is for-profit colleges. These for-profit colleges conduct massive advertisement campaigns, which seems like a fine idea if they provide a quality experience that comes with quality outcomes. However, she mentions that for-profit associate degrees are valued about the same as high school diplomas and below associate degrees from public community college. Since these colleges want to maximize profits, they try to target people and help them take government loans to pay for themselves. This is extremely harmful compounded with the fact that they target poor zip codes, recent mothers, people who had a loved one die recently, and people who are unable to see and plan for the future. That is to name a few of the many demographics that they specifically target.

They target disadvantaged people and make them go into tens of thousands of dollars in debt. This makes them more disadvantaged, and thus the feedback loop has started. This negative feedback loop is a prime component of using math and machine learning models to cause great harm. This is the definition of a WMD, as O’Neil describes, a Weapon of Math Destruction.

Targeted online advertisement has grown up to be massively evil by many entities. In old marketing, getting data like this would take months to years to get, and these models do it automatically. In my opinion, this has too much potential for greedy, immoral capitalists to hurt and exploit people. Therefore, it should not exist.

--

--

Daniel Johnson
0 Followers

Computer Science student at KU. Main interests: Machine Learning and AI