Race, artificial intelligence, and systemic inequalities

Brookings is closely monitoring developments with the global spread of COVID-19. Out of an abundance of caution, we are asking any visitors to Brookings to reschedule their visit, tune in via webcast, or conduct their meeting by phone or videoconference if they or someone they reside with has traveled within the past 14 days to a country designated by the U.S. Centers for Disease Control as “Level 3 – Avoid Nonessential Travel” or “Level 2-Practice Enhanced Precautions.”

Concerns around systemic inequalities continue to pervade conversations around the creation and deployment of artificial intelligence (AI) systems. Discussions around how biases show up in the use of machine learning algorithms have been ongoing, suggesting the need for actions that promote more interdisciplinary collaborations. On the one hand, some biases are baked into algorithmic models, especially those adopting the norms, values and assumptions of developers. On the other hand, some algorithms adapt to the historical contexts in which they are deployed and result in decisions that overtly discriminate or exclude certain populations. As AI becomes more ubiquitous in use cases that include employment, housing, credit and admissions, it’s important to disentangle the extent to which racial biases and other forms of systemic inequalities pose ethical and fairness challenges, along with a set of offenses that potentially violate existing anti-discrimination laws.

On March 24, the Center for Technology Innovation at Brookings will host a panel of distinguished computer and social scientists to talk about the intersection of race, AI and systemic inequalities. The discussion will also pursue strategies for creating more engineering equity in mathematical computations that influence AI models.

After the panel, speakers will take audience questions. This event will be webcast live.