The Technology of Outrage: Bias in Artificial Intelligence

Algorithms are not inherently neutral: The paper challenges the common assumption that machines can be fair and unbiased by nature, demonstrating that this belief "quickly leads to absurdity" when examined critically alongside the claim that algorithms can simply replace human decision-making.

"Bias" means different things in different contexts: The authors identify "equivocation surrounding the word 'bias'" and distinguish between intellectual, moral, and political forms of outrage when people react to algorithmic bias, suggesting that clearer language is essential for productive conversations about AI fairness.

Three practical solutions exist: The paper recommends that the AI community (1) clarify language around bias, (2) develop new auditing methods to test intelligent systems for fairness problems, and (3) build specific capabilities into AI systems to address bias—moving beyond abstract discussions to concrete action.

Emotional reactions to algorithmic bias reveal deeper concerns: The authors analyze why people experience "outrage" when encountering biased AI, suggesting these reactions reflect legitimate intellectual, moral, and political concerns that deserve serious consideration rather than dismissal.

Current approaches to AI fairness are insufficient: The evidence presented indicates that simply replacing human decision-makers with algorithms does not eliminate bias; instead, it can obscure bias and shift responsibility away from human accountability.