AI Safety Considerations for Agents With Limited Time to Act
We discuss theoretical bounds for agent-agnostic safety guarantees in environments that can only be partially observed and within which an action is required within limited time.
ProofPaper ↗
Key points
- In the wake of the increasingly public discussion about AI alignment, recent work has tried to propose specific AI architectures that behave safely.
- We introduce two realistic scenarios, one with an infinite state space and one with signal mixture.
- In these scenarios, we prove that even a perfect agent cannot guarantee safe behaviour.
- It will be argued that for any proof of AI safety or alignment, the environment and associated safe actions need to be specifically considered together with the agent.
Sources (1)
- [1]AI Safety Considerations for Agents With Limited Time to ActarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 03:49 PM
We discuss theoretical bounds for agent-agnostic safety guarantees in environments that can only be partially observed and within which an action is required within limited time.
In the wake of the increasingly public discussion about AI alignment, recent work has tried to propose specific AI architectures that behave safely.
Extractive summary: sentences quoted from the sources.