XAI presentation

Prediction Accuracy

Traceability

Decision Understanding

Explainability

Traceability

DeepLIFT

LIME

Decision Understanding

knowing how the underlying (AI) technology works (rules)

knowing how the AI-based system came up with a given result

The movement of AI models to new platforms

AI models learning how to drift muscle cars

The degradation of model performance over time due to changing data

The improvement of AI models over time without intervention

Decreses the computational cost of AI models

Extracts human-interpretable rules and concepts from black-box models

Simplifies the data collection process

Reduces the need for domain expertise in AI development

Adversarial parties

Responsibility vs Explainability

Understanding versus Trust

Technical complexity