Course Overview
The development of AI has created new opportunities to improve the lives of people around the world, from business to healthcare to education. It has also raised new questions about the best way to build fairness, interpretability, privacy, and safety into these systems.
In this course, you will do a high-level exploration of Google's recommended best practices for responsible AI usage across different areas of focus: Fairness, Interpretability, Privacy and Safety. Along the way, you will learn how you can leverage different open-source tools and tools on Vertex AI to explore these concepts and spend time considering the different challenges that arise with generative AI.
Who should attend
Machine learning practitioners and AI application developers wanting to leverage generative AI in a responsible manner.
Prerequisites
To get the most out of this course, participants should have:
- Familiarity with basic concepts of machine learning
- Familiarity with basic concepts of generative AI on Google Cloud in Vertex AI
Course Objectives
- Overview of Responsible AI principles and practices
- Implement processes to check for unfair biases within machine learning models
- Explore techniques to interpret the behavior of machine learning models in a human-understandable manner
- Create processes that enforce the privacy of sensitive data in machine learning applications
- Understand techniques to ensure safety for generative AI-powered applications
Outline: Introduction to Responsible AI in Practice (IRAP)
Module 1 - AI Principles and Responsible AI
- Google's AI Principles
- Responsible AI practices
- General best practices
Module 2 - Fairness in AI
- Overview in Fairness in AI
- Examples of tools to study fairness of datasets and models
- Lab: Using TensorFlow Data Validation and TensorFlow Model Analysis to Ensure Fairness
Module 3 - Interpretability of AI
- Overview of Interpretability in AI
- Metric selection
- Taxonomy of explainability in ML Models
- Examples of tools to study interpretability
- Lab: Learning Interpretability Tool for Text Summarization
Module 4 - Privacy in ML
- Overview in Privacy in ML
- Data security
- Model security
- Security for Generative AI on Google Cloud
Module 5 - AI Safety
- Overview of AI Safety
- Adversarial testing
- Safety in Gen AI Studio
- Lab: Responsible AI with Gen AI Studio