AI Ethics
Safe Exploration
Safe Exploration is a strategy that ensures discovering new information or opportunities while minimizing risks and negative consequences.
AI Ethics
Safe Reinforcement Learning
Safe Reinforcement Learning is a branch of reinforcement learning focused on ensuring that autonomous agents learn and act without causing harmful or risky behaviors.
AI Ethics
Side Effects
Side effects are unintended or secondary outcomes that occur in addition to the primary intended effect of an action, treatment, or process.
AI Ethics
Society-in-the-loop
Society-in-the-loop is a framework that integrates societal values and ethical considerations directly into the design and governance of AI systems.
AI Ethics
Specification Gaming
Specification gaming is when developers or users exploit loopholes or ambiguities in a set of rules or specifications to achieve desired outcomes, often bypassing the intended purpose.
AI Ethics
Surveillance
Surveillance is the systematic observation and monitoring of people, activities, or information to gather data and ensure security or compliance.
AI Ethics
System Cards
System Cards are structured digital or physical tools used to organize, display, and manage system components or functionalities in a clear and accessible format.