The AI Taxonomy.
Terms  /  System Transparency  /  Accountability Gap
06 · System Transparency

Accountability Gap

No one — human or system — can explain why a specific output was produced.

The Accountability Gap is the ultimate consequence of opacity. When something goes wrong — a bug, a security vulnerability, a broken deployment — and nobody can explain how it happened, you've fallen into the Accountability Gap. The human didn't write the code. The agent can't explain its reasoning after the fact. The system doesn't log its decision process. The result is an outcome with no author, no explanation, and no clear path to prevention. In regulated industries, the Accountability Gap isn't just problematic — it's potentially illegal.

Example
A production system fails because an AI-generated configuration had an incorrect timeout value. The team investigates: the developer didn't set the value (the agent did). The agent's reasoning isn't logged. The prompt that led to the configuration is lost (session expired). The system doesn't record which model version was used or what context was provided. Nobody can explain why the timeout was set to 50ms instead of 5000ms. The post-mortem concludes with "we think the AI made an error" — which is not an explanation, it's an admission of an Accountability Gap.

More in System Transparency

System Opacity

The inability to see what an AI system knows, why it made a decision, or what it did. The default state of most AI tools today.

Decision Fog

The user can see the output but not the reasoning, trade-offs, or alternatives the model considered.

Black Box Agency

An agent that acts on your behalf but won't show its work. You get results with no audit trail.

Invisible State

The system has internal state that affects behavior but is not exposed to the user.

Reasoning Visibility

The user can inspect the model's chain of thought, assumptions, and decision points.

Decision Traceability

Every output can be traced back through the reasoning, context, and data that produced it.

Open State

The system's internal state is visible and inspectable at any time. The opposite of Invisible State.

Audit Trail

A complete, immutable record of what the system did, when, why, and with what inputs.