Bias and Discrimination in AI (edX)

Engage in this course pertaining to a highly impactful yet, too rarely discussed, AI-related topic. You will learn from international experts in the field, also speakers at IVADO’s International School on Bias and Discrimination in AI, which took place in Montreal, and explore the social and technical aspects of bias, discrimination and fairness in machine learning and algorithm design.

The main focus of this course is: gender, race and socioeconomic-based bias as well as bias in data-driven predictive models leading to decisions. The course is primarily intended for professionals and academics with basic knowledge in mathematics and programming, but the rich content will be of great use to whomever uses, or is interested in, AI in any other way. These sociotechnical topics have proven to be great eye-openers for technical professionals!

The total duration of the video content available in this course is 7:30 hours, cut into relevant segments that you may watch at your own pace. There are also comprehensive quizzes at the end of each segment to measure your understanding of the content.

IVADO is a scientific and economic data science hub bridging industrial, academic and governmental partners with expertise in digital intelligence. One of its missions is to contribute to the advancement of digital knowledge and train new generations of bias-aware data scientists.

Welcome to this enlightening journey in the world of ethical AI!

What you'll learn

- Understanding bias and discrimination in all its aspects

- Exploring the harmful effects of bias in machine learning (discriminatory effects of algorithmic decision-making)

- Identifying the sources of bias and discrimination in machine learning

- Mitigating bias in machine learning (strategies for addressing bias)

- Recommendations to guide the ethical development and evaluation of algorithms

Syllabus

Module 1 The concepts of bias and fairness in AI

- Different Types of Bias

- Fairness criteria and metrics

Module 2 Fields where problems were diagnosed

- Privacy, labour and legal system

- Public policy and Health

Module 3 Institutional attempts to mitigate bias and discrimination in AI

- Canada's Algorithmic Impact Assessment Framework

- The Montreal Declaration for Responsible AI

Module 4 Technical attempts to mitigate bias and discrimination in AI

- Fairness constraints in graph embeddings

- Gender bias in text