Algorithmic discrimination causes less moral outrage than human discrimination | JoVE Visualize

Area of Science:

Social Psychology
Computer Science Ethics
Algorithmic Bias

Background:

Algorithms are increasingly used in decision-making across various sectors, including hiring, medical treatment, and parole.
While algorithms can potentially mitigate human bias, they frequently perpetuate or introduce new forms of discrimination.
Public perception of algorithmic discrimination is often assumed to be one of moral outrage, similar to human discrimination.

Purpose of the Study:

To investigate the hypothesis of an 'algorithmic outrage deficit,' where people exhibit less moral outrage towards algorithmic discrimination compared to human discrimination.
To examine whether this deficit extends to reduced organizational responsibility and legal liability when discrimination stems from algorithms.
To explore the impact of algorithmic decision-making on reputational benefits and collective action against systemic discrimination.

Main Methods:

Eight studies were conducted using diverse participant groups, including online samples, a quasi-representative sample, and tech workers.
The studies focused on gender discrimination in hiring practices to test the algorithmic outrage deficit hypothesis.
Data collection involved assessing participants' moral outrage, attributions of intent, and judgments of organizational responsibility and legal liability.

Main Results:

Participants demonstrated significantly less moral outrage towards algorithmic discrimination than human discrimination.
Organizations were held less responsible and faced reduced legal liability when discrimination was caused by algorithms versus human employees.
The reduced attribution of prejudicial motivation to algorithms was identified as a key driver of the algorithmic outrage deficit.
Companies received less reputational benefit when algorithms improved gender equality compared to when employees did.

Conclusions:

An 'algorithmic outrage deficit' exists, leading to diminished moral outrage, reduced accountability, and potentially weaker collective action against algorithmic discrimination.
The perception of algorithms as lacking intent reduces blame, impacting societal responses to and regulation of discriminatory AI systems.
These findings have significant implications for understanding public perception of AI ethics and for developing strategies to ensure fairness and accountability in algorithmic decision-making.