Accountability
A system or org taking responsibility for how their AI behaves and why it decided something.
Think of it like
Like a company putting its name on a product — if it breaks, you know who to call.
Example
When a hospital's AI misdiagnoses a patient, someone needs to answer for it, explain the reasoning, and fix the process.
How it actually works
Accountability means traceability and answerability — if the model made a decision, you can explain it and trace back why. It's not just about blame; it's about having clear ownership, audit trails, and mechanisms to course-correct. Hard part: when a model's decision emerges from millions of parameters, it can be genuinely opaque.
For product teams
Build decision trails and escalation paths so stakeholders can understand and contest AI outputs.
For engineers
Log decision paths, feature importance, and model versions so you can reproduce and debug any output.
Related
- Related to traceability and interpretability.
- Works alongside oversight mechanisms.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome