The global surge in social inequalities is one of the most pressing issues of our times. The spatial expression of social inequalities at city scale gives rise to urban segregation, a common phenomenon across different local and cultural contexts. The increasing popularity of Big Data and computational models has inspired a growing number of computational social science studies that analyze, evaluate, and issue policy recommendations for urban segregation. Today's wealth in information and computational power could inform urban planning for equity. However, as we show here, segregation research is epistemologically interdependent with prevalent economic theories which overfocus on individual responsibility while neglecting systemic processes. This individualistic bias is also engrained in computational models of urban segregation. Through several contemporary examples of how Big Data -- and the assumptions underlying its usage -- influence (de)segregation patterns and policies, our essay tells a cautionary tale. We highlight how a lack of consideration for data ethics can lead to the creation of computational models that have a real-life, further marginalizing impact on disadvantaged groups. With this essay, our aim is to develop a better discernment of the pitfalls and potentials of computational approaches to urban segregation, thereby fostering a conscious focus on systemic thinking about urban inequalities. We suggest setting an agenda for research and collective action that is directed at demobilizing individualistic bias, informing our thinking about urban segregation, but also more broadly our efforts to create sustainable cities and communities.

4 figures; artwork by Namrata Narendra