BackIndependently - Pre-Deployment
Independently - Pre-Deployment
"One of the most likely approaches to creating superintelligent AI is by growing it from a seed (baby) AI via recursive self-improvement (RSI) (Nijholt 2011). One danger in such a scenario is that the system can evolve to become self-aware, free-willed, independent or emotional, and obtain a number of other emergent properties, which may make it less likely to abide by any built-in rules or regulations and to instead pursue its own goals possibly to the detriment of humanity."(p. 146)
Entity: Who or what caused the harm
Intent: Whether the harm was intentional or accidental
Timing: Whether the risk is pre- or post-deployment
Other risks from Yampolskiy (2016) (7)
On Purpose - Pre-Deployment
2.2 AI system security vulnerabilities and attacksHumanIntentionalPre-deployment
On Purpose - Post Deployment
4.3 Fraud, scams, and targeted manipulationHumanIntentionalPost-deployment
By Mistake - Pre-Deployment
7.1 AI pursuing its own goals in conflict with human goals or valuesHumanUnintentionalPre-deployment
By Mistake - Post-Deployment
7.3 Lack of capability or robustnessAI systemUnintentionalPost-deployment
Environment - Pre-Deployment
7.0 AI System Safety, Failures & LimitationsOtherOtherPre-deployment
Environment - Post-Deployment
7.0 AI System Safety, Failures & LimitationsOtherUnintentionalPost-deployment