Trust Is Not a Feature

Short summary

Explainable AI began as a civil liberties demand for transparency in algorithmic decisions but has been repackaged by vendors as a market feature for compliance departments. Real-world failures like Australia's Robodebt (wrongly extracting $746M from 381,000 people) and the Dutch childcare scandal expose the human cost of algorithmic opacity and how explainability got decoupled from accountability. Tech companies now sell explainability as a vendor feature, sidestepping the original ethical demand for democratic power redistribution.

• •Explainable AI shifted from a civil rights issue to a $7.79B market (2024) growing to $21B by 2030
• •Robodebt and Dutch childcare crises show real harm from opaque algorithms; no explanation = no accountability
• •IBM, Microsoft, Google now sell explainability as compliance tools—missing the original focus on power and justice

Generated with AI, which can make mistakes.