Practical Steps for Building Fair AI Algorithms

Algorithms increasingly help make high-stakes decisions in healthcare, criminal justice, hiring, and other important areas. This makes it essential that these algorithms be fair, but recent years have shown the many ways algorithms can have biases by age, gender, nationality, race, and other attributes. This course will teach you ten practical principles for designing fair algorithms. It will emphasize real-world relevance via concrete takeaways from case studies of modern algorithms, including those in criminal justice, healthcare, and large language models like ChatGPT. You will come away with an understanding of the basic rules to follow when trying to design fair algorithms, and assess algorithms for fairness. This course is aimed at a broad audience of students in high school or above who are interested in computer science and algorithm design. It will not require you to write code, and relevant computer science concepts will be explained at the beginning of the course. The course is designed to be useful to engineers and data scientists interested in building fair algorithms; policy-makers and managers interested in assessing algorithms for fairness; and all citizens of a society increasingly shaped by algorithmic decision-making.

• Introduction
• In this module, you'll learn the basic concepts this course relies on: what an algorithm is, and why fairness is tricky and subtle to define. We'll start by defining what a predictive algorithm even is, because this course is designed to be accessible to students who have never taken a computer science class. (If you have taken a previous class on predictive algorithms or machine learning, feel free to skip this section.) Then we'll jump right into fairness. This course will present ten practical fairness lessons, and in this module we'll discuss two of them. We'll also give a sneak preview of how the lessons of this course apply to generative AI models like ChatGPT.
• Designing Algorithms
• This module will cover fundamental lessons for designing fair algorithms: what data they should be trained on, what features they should use to predict, and what outcomes they should predict.
• Documenting Algorithms
• This module discusses the importance of documenting algorithms and datasets so they are used only in settings where they are appropriate.
• Algorithms in the hands of humans
• This module discusses the complex interplay between algorithmic predictions and human decisions.