Knowledge Maxxing← Back to topic library
Responsible Tech MaxxingIntermediateRisk Management75 sec
AI risk
AI risk is the potential for an AI system to cause harm, evaluated in terms of both how likely that harm is and how severe it would be if it occurred.
What matters
Treating risk as a combination of likelihood and severity, rather than either alone, is what lets teams prioritize which risks genuinely need attention first.
Example
A rare but catastrophic failure and a common but minor annoyance are both AI risks, but they warrant very different levels of mitigation effort.
Common misconception
AI risk isn't certainty of harm; a risk can be worth taking seriously even if the harmful outcome is unlikely, if its severity would be great enough.
Recall question
What two factors combine to determine how seriously an AI risk should be treated?
Keywords and related concepts
AI riskharm potentialrisk assessmentlikelihood