Practice question
A government agency plans an AI eligibility system. Testing shows systematic disadvantage to a linguistic minority. The team refuses bias audits because aggregate accuracy is high and tells applicants that proprietary complexity makes any explanation impossible. Under section 5 of the issued DICT–CSC JMC No. 003, s. 2026, which response best addresses the principles?
- High average accuracy excuses discriminatory effects
- Fairness requires bias detection and mitigation, while transparency requires applicable plain-language explanations of relevant factors, risks, limitations, and dispute mechanisms
- A proprietary model is exempt from every general principle
- Applicants must accept outputs because government AI is necessarily infallible
- Only environmental impact matters once software is procured
Legal references
- https://elibrary.judiciary.gov.ph/thebookshelf/showdocs/11/101138 — A.M. No. 25-11-28-SC, February 18, 2026, AI Governance Framework, Part II, ethical governance principles.
- https://www.officialgazette.gov.ph/downloads/2026/06jun/20260609-DICT-CSC-JMC-003-2026-FRM.pdf — DICT–CSC JMC No. 003, s. 2026, issued June 9, 2026, section 5; section 15 for effectivity conditions.
Answer this question in flashcards to review your response and continue practicing.