Sweating Bullets — AI and Workers’ Lives

Posted on April 7th, 2021 by Emilio Miles

In chapter 7 of Weapons of Math Destruction, author Cathy O’Neil talks about the misuse of scheduling software and teacher evaluation software and how it affects the lives of people. She states that they create a poisonous feedback loop for workers that ends up causing more hurt than good. Like most WMDs she speaks of in the book, the software discussed in this chapter is erroneous in the way it measures its own success.

Scheduling software is used to maximize efficiency in the workplace while minimizing labor costs. More people are scheduled during busy hours and labor is cut when it is slow in the stores. Scheduling software targets people who would be more likely to need money and therefore more willing to work on short notice. O’Neil talks about the case of Janette Navarro and how Starbucks, embarrassed by the blowback, made the “rule” to post all work hours at least one week in advance. However, that didn’t last long and before long, Starbucks was back to springing hours on their employees on short notice. It was easier to manage the labor budget by just changing someone’s schedule than it was to upkeep the company pledge.

Scheduling software created a poisonous feedback loop for Janette Navarro by making it impossible for her to return to school due to her haphazard scheduling. This dampened her employment prospects and kept her in the pool of low-wage workers. The long and irregular hours also made it difficult for her, and other workers, to organize and protest for better conditions. These chaotic schedules also make it almost impossible to find a second job and thus keeps them in the same job, keeping them in that loop.

Tim Clifford was another person affected by WMDs. He was an English teacher in New York City who was targeted by the use of teacher evaluation software. One year, he received a score of 6 out of 100, which would have cost him his job were it not for tenure. He was angry, confused, and worried. The evaluation software didn’t tell him what he was doing wrong so he had no way of telling how he could improve. The following year, however, he taught the exact same way and received a whopping 96 out of 100. It turned out to be that it was because of the type of students he had in his class. The year prior, he had mostly poor students who performed poorly or really smart students who always performed great and had little room for improvement. This made the software “see” that he had not made a significant impact on his students since they were not improving.

However, the following year, he had a more diverse group and thus his score saw a bigger change. More students were able to raise their grade. This revealed the major flaw in these evaluation programs. They oversimplified the performance of a teacher which often caused great teachers to lose their job because of faulty algorithms. Programs like this should be done away with or reformed.

In recent years, universities have started making test scores optional for admissions. Is this a good idea? It can be for some students. If you are a student with a great GPA in high school, involved in many extracurriculars, and have a great “activity list,” but you tested poorly, for whatever reason, then this is great solution. You get evaluated on your grade trends rather than one test score. The option to have your test scores reviewed, or not, when applying for college admission then seems like common sense. Students should have the ability to choose whichever option would maximize their chances of receiving higher education.

Computer Science student at the University of Kansas