Advocates of algorithmic correction suggest taking lessons from management professionals such as librarians, who have had to consider how to ethically collect data about people and what should be done. put in the library. They propose not only to consider the performance of an AI model as fair or good, but also whether it changes power.
Proposals echo previous suggestions from former Google AI researcher Timnit Gebru, ai in a newspaper in 2019 mobilize practitioners of machine learning to examine how archivists and library scientists deal with issues of ethics, inclusivity, and authority. Gebru says Google fired her at the end of 2020 and recently launched a distributed AI research center. One important analysis concluded that Google forced Gebru into a historical pattern of abuse targeting Black women in a professional setting. The authors of that analysis also urge computer scientists to look for historical and social patterns beyond the data.
Earlier this year, five US senators urge Google Hire an independent auditor to assess the impact of racism on Google products and the workplace. Google did not respond to the letter.
In 2019, four Google AI researchers Discuss The responsible AI field needs significant race theory because most work in the field doesn’t take into account the built-in social dimension of race or recognize the influence of history on data sets. collect.
“We emphasize that data collection and annotation efforts must be based on the social and historical context of racial classification and racial formation,” the paper reads. “Oversimplification is the practice of violence, or even more, the depiction of violence against communities that have experienced structural violence.”
Lead author Alex Hanna was one of the first sociologists hired by Google and the lead author of the paper. She was a fierce critic of Google executives after Gebru’s departure. Hanna said she appreciate that critical racial theory focuses on conversations about what is fair or ethical and can help reveal historical patterns of oppression. Since then, Hanna co-authored an article to be also published in Big Data & Society How does that confront? face recognition technology reinforces gender and racial structures dating back to colonial times.
At the end of 2020, Margaret Mitchell, who with Gebru leads the Ethical AI team at Google, speak Companies have begun to use critical race theory to help decide what is fair or ethical. Mitchell was fired in February. A Google spokesperson said race theory is important as part of the AI research review process.
Again paper, policy adviser to the White House Office of Science and Technology Rashida Richardson, to be published next year, asserts that you cannot think of AI in America without acknowledging the effects of racism. The legacy of laws and social norms to control, exclude, and oppress Negroes was overwhelming.
For example, studies have found that algorithms used to apartment rental screen and mortgage applicant disadvantage disadvantage Black people. Richardson says it’s important to remember that federal housing policy explicitly required racial segregation until civil rights laws were passed in the 1960s. The government also colluded with developers. and landlords to deny opportunities to people of color and separate racial groups. She says segregation has triggered “cartel-like behavior” among whites in homeowners’ unions, school boards and unions. In turn, detached housing takes on complex or privileged issues related to the education or wealth of the generation.
Richardson says that discriminatory patterns have historically tainted the data on which many algorithms are built, such as to classify what is a “good” school or attitudes about controlling neighborhoods. Brown and Black.
“Racism has played a central evolutionary role in recreating and amplifying racial stratification in data-driven technologies and applications. Racism also limits conceptualization of algorithmic bias issues and related interventions,” she wrote. “When the effects of racism are ignored, problems of racial inequality emerge as a natural phenomenon, rather than a by-product of policies, practices, specific social norms and behaviors”.
