ExplainabilityTransparencyTrustworthinesscs.AI
TrustyAI Explainability Toolkit
Rob Geada, Tommaso Teofili, Rui Vieira, Rebecca Whitworth, Daniele Zonca
April 26, 2021
arXiv.org
3 citations
Key Takeaways
-
AI systems are often "black boxes" that are too complex to understand, making it difficult for organizations and individuals to know why these systems make specific decisions—a critical problem as AI becomes more prevalent in workplaces and homes.
-
The TrustyAI Explainability Toolkit provides practical tools to make AI decisions more transparent by offering Java and Python libraries that can explain both AI models and decision services, making explainability accessible to both enterprise users and data scientists.
-
Multiple explanation techniques are now available and standardized, including LIME, SHAP, and counterfactual methods, which the toolkit implements and benchmarks—giving organizations options for understanding their AI systems depending on their specific needs.
-
Explainability solutions are essential for responsible AI deployment, as understanding why an AI system makes a decision is fundamental to ensuring it behaves as intended and to building appropriate trust in automated decision-making systems.
Abstract
Artificial intelligence (AI) is becoming increasingly more popular and can be found in workplaces and homes around the world. The decisions made by such "black box" systems are often opaque; that is, so complex as to be functionally impossible to understand. How do we ensure that these systems are behaving as desired? TrustyAI is an initiative which looks into explainable artificial intelligence (XAI) solutions to address this issue of explainability in the context of both AI models and decision services. This paper presents the TrustyAI Explainability Toolkit, a Java and Python library that provides XAI explanations of decision services and predictive models for both enterprise and data science use-cases. We describe the TrustyAI implementations and extensions to techniques such as LIME, SHAP and counterfactuals, which are benchmarked against existing implementations in a variety of experiments.