Big decisions about people’s lives are increasingly made by software systems and algorithms. Sorting résumés for job applications, allocating social services, and deciding who sees advertisements for open positions, housing, and products are just a few of the ways in which these software systems shape our lives.
While algorithmic decision-making can offer benefits in terms of speed, efficiency, and even fairness, bias is routinely introduced into software systems in many ways, including the use of biased training data.
Often “black boxes” with little transparency or accountability, algorithms can unfairly limit opportunities, restrict services, and produce “technological redlining”–a form of digital data discrimination that uses digital identities and activities to bolster inequality and oppression.
Algorithmic Accountability: A Primer explores issues of algorithmic accountability, or the process of assigning responsibility for harm when algorithmic decision-making results in discriminatory and inequitable outcomes.
Currently, there are few consumer or civil rights protections that limit the types of data used to build data profiles or that require the auditing of algorithmic decision-making, even though algorithmic systems can make decisions on the basis of protected attributes like race, income,or gender–even when those attributes are not referenced explicitly–because there are many effective proxies for the same information.
This brief explores the trade-offs between and debates about algorithms and accountability across several key ethical dimensions, including: