AI Now, a research institute that examines the social impact of artificial intelligence, has just published a study in which it diagnoses the AI industry with a "diversity crisis". With this diagnosis, the study addresses questions that we probably had not asked ourselves before when thinking about the development of these new technologies.
The study itself is called "Discriminating Systems" and is available for reading in its original version at this link. According to the findings of the study, several gaps or deficiencies in the field of artificial intelligence are reported. Next is one of them:
Crisis of gender diversity and race in the AI sector: "inequality in the AI industry is extreme". Eighty percent of AI teachers are men, and only 18 percent of the authors at relevant conferences in the field are women. Facebook's AI research department is 15% women and Google's only 10%. There are no public data on the occupation of trans persons or other gender minorities. For people of color, the situation is even worse. For example: only 2.5% of Google workers are nonwhite, while in Facebook and Microsoft the figure reaches only 4%. Given the decades of work and investment to reverse this type of inequality, the situation is alarming.
Of course, the report makes a number of suggestions for improving the current state. For example: companies could improve their transparency by publishing reports on work and its financial compensation, broken down by race and gender. The publication of transparency reports on harassment and discrimination is also suggested.
Well, even with the exposition of these problems and suggestions, the approach remains in the field of abstraction, but has very clear consequences in daily life and in the daily use of various technologies. For example, the unconscious biases of white men may influence achievements designed for facial recognition, in turn affecting historically marginalized groups. To name an example, the use of a program to guess people's sexual orientation using a facial recognition system, an experiment carried out by researchers at Stanford University.
Many workers in the technology industry have risen to point out major problems in the development of artificial intelligence, prompting the companies in which they work to suspend or revise the use of tools with the potential to harm vulnerable or minority groups. Amazon employees have questioned managers about the company's use of facial recognition. More recently, Google employees rose up against an IA ethics oversight board that included the president of the Heritage Foundation, a group known for lobbying against the rights of LGBTG people. In response, the company dissolved the board completely.
The authors of the report conclude that the crisis of diversity in the field of AI is well documented and very broad in scope. "It can be seen in unequal workplaces throughout industry and in academia, in the disparities in hiring and promotion, in the AI technologies that reflect and amplify biased stereotypes, and in the resurfacing of biological determinism in automated systems."
What do you think about that? Have any of you experienced or witnessed a case of discrimination by a system based on artificial intelligence? You can tell us about your experience in the comments.