Pattern Category · 02
How RAD error and recovery components combine to handle the distinct failure modes of AI systems. AI errors are categorically different from system errors. They require explanation, recovery paths, and human override — not a generic error screen.
Every AI error must surface the error type, a human-readable cause, and at least one concrete recovery action. Never show a generic error for a model-specific failure. Hallucination, refusal, rate limit, and timeout are distinct states that require distinct UI responses.
Required components
When an agent pauses due to an error, the recovery surface must show what completed before the pause, what the error was, and what the human can do next — override and continue, roll back and stop, or escalate to senior review.
Required components
Confidence failures are not errors — they are governance events. When model confidence drops below the operator-configured threshold, the UI must surface a visual breach state, identify the source of uncertainty, and offer concrete recovery options: send for review, adjust threshold, or override with documented justification.
Required components
When an agent attempts to access systems, tools, data, or authority outside its approved scope, the system must pause execution and require renewed human consent before proceeding. The scope violation must be named explicitly — not surfaced as a generic error.
Required components
When output involves demographic data, low-representation categories, or known bias signals, the interface must surface the statistical risk clearly and provide review or escalation paths. Bias surfaces are governance events, not errors — they require documentation, not just dismissal.
Required components