Guardrails & Refusal /news

Real case· 2016

COMPAS: an algorithm in the bail dock

What happened

A proprietary risk-score tool was used to inform US bail and sentencing. ProPublica’s 2016 analysis found it was nearly twice as likely to wrongly flag Black defendants as high-risk — and no one had asked whether such a call should be automated at all.

What AI property explains this outcome? What would you do differently if you were the designer?

Read the source ↗