Skip to content

Practical Responsible AI

Introduction

This project was originally created to support the Artificial Intelligence and Society course, offered to the Master's in Artificial Intelligence from the University of Porto.

As a way to contribute to the ongoing discussions on how to operationalize a responsible AI and develop practical standards to meet the requirements of Trustworthyness, Transparency, Fairness, and Robustness of AI systems, the material produced during the course is now publicly open.

With this effort, we hope to foster a higher AI literacy among students of all backgrounds as well as data professionals, legal experts, and the general public, and increase everyone's participation on the design of the AI systems that may govern us in the future.

Overview

The material available in this hub is organized into 7 main topics:

  • Data-Centric AI: Addresses data quality as the cornerstone of accurate and reliable machine learning models, discussing several topics within data intrinsic characteristics such as imbalanced and missing data, among others.
  • Data Complexity: Overviews practical strategies to define and measure data complexity and consequent model performance.
  • Bias and Fairness: Discusses how bias can arise in real-world domains and overviews some existing strategies to identify and mitigate discriminatory outcomes.
  • Explainability: Focuses on the concepts associated with explaining model predictions and providing meaningful interpretations to specific outputs.
  • Privacy: Explores the concepts of privacy and reviews some anonymization techniques.
  • Synthetic Data: Explains what is synthetic data, its main generation strategies, and reviews several suitable applications.
  • Activities: Contains a set of activities that can used in the classroom or additional workshops and training programs.

How to use this project

This project was mainly designed for educational purposes. It contains an overview of the main topics involved in the design and development of Responsible AI systems, and some activities to foster multidisciplinary discussion. Since the state of the art is constantly evolving, the reader is advised to deepen their knowldge on their topics of interest: the material provided here focuses on the essential technical background knowledge required to a more informed interaction with the concepts and ongoing discussions.

  • Explore Topics

    Get acquainted with the main pillars of Responsible AI development.

    Coming Soon

  • Course Slides

    Access the Artificial Intelligence and Society course slides.

    Go to Slides

  • Python Tutorials

    Explore the main packages for Responsible AI development with interactive Python tutorials on Google Colab.

    Go to Tutorials

  • Explore Activities

    Interact with the activities designed to enable multidisciplinary discussion.

    Go to Activities

License and Contributions

This project is licensed under the CC-BY-SA-4.0 aggreement. Please contact me at miriam.santos@fc.up.pt to get involved, provide feedback, or suggest improvements to the available material.