Technological Justice

I watched a documentary on Netflix called Coded Bias recently. It mainly followed the journey of Dr. Joy Buolamwini as she worked to expose some of the deficiencies of artificial intelligence (AI). As articulated by Dr. Buolamwini some of the problems of AI include its general inability to accurately identify women and people of color via facial recognition algorithms and its replication of human biases within algorithms.

I have read about both of these issues in the past. While watching this documentary and seeing the intrusion of these issues into the world of the mundane I was struck by the feeling that some of these issues are deliberate. When I have read about these issues in the past they have always been glossed over as accidental.

What if that isn’t true? In the United States companies do not generally pay computer programmers to code for social justice. The point of an organization hiring computer programmers in a free market economy is to increase efficiencies in their operations thereby leading to increased profits. If an organization uses a business model that is dependent on exploiting employees and consumers then that organization would use a computer programmer to exploit employees and consumers more effectively. There is no upside to that organization suddenly erring on the side of morality or justice. In such a case, morality and justice are a drag on profits.

It is logical that a for profit company would use technology to improve their ability to exploit people. When we read about racial and gender bias in technology it is typically attributed to human error. There are always promises to make improvements. However, another point made in this documentary was about the opacity of the processes using algorithms. The employees and consumers being exploited rarely have any oversight over, or even knowledge of, any corrections being made.