Trustworthy AI Characteristics
Seven interrelated characteristics that together constitute AI trustworthiness:
- Valid and Reliable: System performs as required under expected conditions; accuracy and robustness are demonstrated
- Safe: System does not endanger human life, health, property, or environment under defined conditions
- Secure and Resilient: System maintains confidentiality, integrity, and availability; withstands adverse events
- Accountable and Transparent: Information about system and outputs is available to those interacting with it
- Explainable and Interpretable: Mechanisms underlying operation can be represented; outputs have meaning in context
- Privacy-Enhanced: Practices safeguard human autonomy, identity, and dignity
- Fair with Harmful Bias Managed: Systemic, computational, and human-cognitive biases are identified and mitigated
Valid & Reliable is positioned as foundational, other characteristics build upon it. Accountable & Transparent relates to all other characteristics.
Creating trustworthy AI requires balancing each characteristic based on context of use. Tradeoffs are unavoidable.