Explainable AI

Quick Definition

AI that provides understandable explanations for decisions

AI systems designed to provide human-understandable explanations for their decisions.

medical diagnosis justification, credit scoring

interpretability ai-ethics

Training models to predict human preferences for RLHF

Neural network with cyclic connections for sequence processing

Text model representing documents by word frequency counts

Removing redundant neural network connections to reduce size

Creating smaller models that replicate larger model behavior

Connecting AI outputs to verifiable real-world sources