Ethical ML - Who's Afraid of the Black Box Models

Dive into the ethical challenges of explainable machine learning in this conference talk from GOTO Copenhagen 2021. Explore the pitfalls of interpretability and explainability in AI models, and learn how ethical data handling and counter-factual fairness model testing can satisfy GDPR and ALTAI guidelines. Follow Principal Data Scientist Prayson Daniel as he discusses the importance of responsible AI development, acknowledges past mistakes in the field, and demonstrates practical solutions. Gain insights into balancing the complexity of black-box models with regulatory compliance and ethical considerations. Through a comprehensive demo and thought-provoking conclusions, discover strategies for developing trustworthy AI systems that prioritize fairness and transparency.

Intro
Dragons & unicorns
Beyond model transparency & explainability
Why should you care?
We data scientists have messed things up
We data scientists are fixing it
Demo
Conclusions
Outro