Responsible AI
Responsible AI is the practice of developing and operating AI with attention to its effects on people, including fairness, privacy, safety, transparency and accountability. It turns those concerns into decisions, checks and assigned responsibilities across a system’s lifecycle.
Before using a hiring assistant, a team could document its intended role, assess privacy and fairness risks, measure relevant errors across groups and assign a person to review adverse recommendations. After launch, the team needs a way to investigate problems and change or withdraw the system. These are illustrative practices, not a compliance checklist.
NIST’s AI Risk Management Framework organizes work into govern, map, measure and manage. It treats reliability, safety, security, transparency, explainability, privacy and fairness as related qualities that depend on context and can involve tradeoffs.
The term overlaps with AI governance, which establishes organizational controls, and AI safety, which focuses on reducing harm. There is no single responsible-AI score that proves an application is acceptable. Name the affected people, foreseeable failures and decision owner; a policy statement alone does not show those risks are managed.
Sources
- NIST: AI Risk Management Framework 1.0 — Defines trustworthiness characteristics, lifecycle risk management and govern/map/measure/manage functions.
Go deeper
- NIST: AI RMF Playbook docs
Explore suggested actions organized around the framework’s four risk-management functions.