Algorithms can perpetuate discrimination.
Voice
The frequent assignment of more feminine voices to virtual personal assistant features is not coincidental. Even when more options to the default voice for the virtual assistant are offered, differences between the more masculine and the more feminine voices include not only changes in vocal tones and inflections, but in their responses to certain questions. These differences in representations of gender through artificial intelligence and technology reflect issues of inequality in our realities outside of technology.
Facial Recognition
When Joy Buolamwini discovered that her darker complexion was less identifiable to facial analysis software than her face with a white mask over it, the exclusive tendency of artificial intelligence became especially apparent. The error rates of facial recognition technology by top tier companies are much higher for those of darker complexions, especially women of darker complexions (error is no more than 1% for men of lighter complexions, while error for women of darker complexions is roughly 35%).