| Summary: | <p>Most definitions of algorithmic bias and fairness encode decisionmaker interests, such as profits, rather than the interests of disadvantaged groups (e.g., racial minorities). Bias is equated to a deviation from profit maximization. Future research should instead focus on the causal effect of automated decisions on the distribution of welfare, across and within groups.</p>
<p>The literature emphasizes the apparent contradictions between different notions of fairness and profit motives. These contradictions vanish when profits are approximately maximized. Existing work involves conceptual slippages between statistical notions of bias and misclassification errors, economic notions of profit, and normative notions of bias and fairness.</p>
<p>Notions of bias nonetheless carry some interest within the welfare paradigm, if we understand bias and discrimination as mechanisms and potential points of intervention.</p>
|
|---|