How algorithms lead to pay discrimination, sexism and racism
A professor at the University of California unveiled how large corporations employ neural networks to keep their employees in check
AI can suggest which Amazon couriers and Uber drivers can be paid less to work more
The algorithm may discriminate against women in the hiring process, believing that they do not need a job
The LAPD used the PredPol algorithm which turned out to be racist
Veena Dubal, a professor at the Hastings College of Law at the University of California, has uncovered how large corporations utilise neural networks to exert control over their employees.
According to Dubal’s research, companies like Lyft, Uber, and Amazon employ algorithms that analyse a vast array of data on employee behaviour and salary. These algorithms then suggest ways to “optimise” processes that ultimately infringe upon employees’ rights.
For instance, if a delivery person or taxi driver earns $100 per day, the AI may recommend reducing their pay rate for each order, effectively pressuring the employee to work longer hours. Dubal highlights that this AI-driven approach shapes employee behaviour in favour of the corporations, referring to it as “algorithmic pay discrimination.”
Artificial intelligence can also perpetuate discrimination in other domains. Between 2014 and 2017, Amazon utilised algorithms in its hiring process, but it soon became evident that the AI exhibited bias against women. As the algorithm was trained on resumes of existing employees, the majority of whom were male, it concluded that women were less desirable for Amazon. Consequently, the company discontinued the use of this tool.
In 2016, researchers at the nonprofit Human Rights Data Analysis Group also accused AI of discriminatory practices. During that time, the LAPD employed the PredPol algorithm, designed to identify areas where crimes were likely to occur in the future.
The researchers discovered that the algorithm directed patrols predominantly to African-American neighbourhoods, where the police had previously been active. However, statistics indicated that drug-related crimes were more prevalent in predominantly “white” areas that received less police attention.
