Today I want to delve into some of the society implications of using Big Data. Does Big Data exclude certain groups of people? Big Data is generated by, about and for people which introduces an element of exclusions of certain segments of the population from the data. Because humans are involved in the interaction with Big Data, there will always most likely be an unintentional bias. According to Barocas’ article “Big Data’s Disparate Impact”, ‘without care, data mining can reproduce existing patterns of discrimination, inherit the prejudice of prior decisionmakers, or simply reflect the widespread biases that persist in society (674). For example, classes and target data specifically choose which variables to focus on which could create discrimination. Subjective data metrics are defined by the analyst or another human making it an imperfect discipline.
Statistics return records to a specific query whereas data mining determines relationships (Barocas 677). Any data that falls outside of standards require human judgement and thus are imperfect in including all segments of society (681). According to the article in the British Medical Journal, a new database replicated existing prejudiced data in its admissions database. St. George’s Hospital in the past had admitted less minority candidates. This old biased data affected the new database and algorithms and models associated with it. Being quantitative doesn’t protect against bias and I think we’ll see more potential for Big Data to exclude as this field grows.
Algorithm criteria are often unclear or difficult to uncover, which means it’s difficult to know to what extent and in what ways the algorithm is biased (Gillespie 10). For example, Twitter hides its evaluative criteria for reporting to users what terms are “trending” in their area at any given time. Twitter and other social media platforms also unintentionally exclude segments of the population without smartphones or are uncomfortable using new technologies, including the poor and elderly. Algorithms by design create groups and boundaries about people which has great potential but also can be harmful (Gillespie 23 and Crawford 2).