Defining fairness
Practice problem
Text 1
A computer ethicist argues that talk of fairness in automated decision systems becomes empty without explicit definitions. They contend that organizations should adopt clear, provisional definitions (for example, committing to a particular standard such as equal false-positive rates across groups) and then revise them in public as evidence accumulates. While no single definition will fit every context, the ethicist warns that refusing to define fairness invites ad hoc judgment, enables selective enforcement, and weakens accountability. “We’ll know it when we see it,” they write, is not clarity but a way to delay responsibility.
Text 2
A regulatory board announces that because contexts vary widely, it will avoid fixing any formal definition of fairness for algorithms. Instead, for each deployment, a rotating panel will assess harms and benefits and issue a case-by-case determination. The board cautions that codifying a definition could ossify standards and stifle innovation; it prefers flexible expert judgment to any metric or formula.
Based on the texts, how would the author of Text 1 most likely respond to the approach described in Text 2?