Results for "long context"
Existential Risk
Advanced
Risk threatening humanity’s survival.
Takeoff Speed
Advanced
Rate at which AI capabilities improve.
Alignment Tax
Advanced
Tradeoff between safety and performance.
AI Boxing
Advanced
Isolating AI systems.
Shutdown Problem
Advanced
Ensuring AI allows shutdown.
Power-Seeking Behavior
Advanced
Tendency to gain control/resources.
Instrumental Goals
Advanced
Goals useful regardless of final objective.
Compute Governance
Intermediate
Regulating access to large-scale compute.
Model Release Control
Intermediate
Restricting distribution of powerful models.