Topic: Bias and Fairness in AI

Lecture

Key Terms:

Algorithmic bias, intersectionality, gender classification, computer vision

Guiding Questions

1. How do biases in AI systems arise and what are their implications?
2. What is intersectionality, and how does it relate to AI ethics?
3. How can we mitigate biases in AI systems?

To Read

"Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification" - Timnit Gebru, Joy Buolamwini

To Read In Class

A Principled AI Discussion in Asilomar - Future of Life Institute

Read all about the 2017 formation of the AI Asilomar Principles, which have since become one of the most influential sets of AI governance principles.

futureoflife.org

AI Principles - Future of Life Institute

At a 2017 FLI conference, AI scientists and researchers developed the highly influential Asilomar AI governance principles. Add your signature.

futureoflife.org

Mission, Team and Story - The Algorithmic Justice League

Fight algorithmic bias with us. We want the world to remember that who codes matters, how we code matters, and that we can code a better future.

www.ajl.org

To Watch

Student Facilitation

Untitled for facilitation schedule

Reading: "Groundwork of the Metaphysics of Morals" - Immanuel Kant

Read Chapter 8 from Ethics: The Essential Writings

Summary: This reading introduces Kant's deontological ethics, which emphasize duty, rules, and universal moral laws. Students will discuss how these principles can be adapted for AI systems, especially around the theme of adherence to ethical guidelines.

Key Questions

1. How does Kant’s concept of duty apply to the development and implementation of AI?
2. How might Kant's categorical imperative guide the ethical programming of AI?
3. How can we interpret the idea of treating humanity "as an end in itself and never as a means to an end" in the context of AI’s interaction with humans?
4. What would it mean for an AI to act out of "duty" and is it even possible according to Kantian ethics?
5. How can Kant’s deontological ethics influence the legal and moral responsibilities of AI and its developers?

To Do

1. Visit the AI Justice League website (https://www.ajl.org/about) individually or in small groups.
2. Explore the website's content, paying attention to the organization's mission, values, and initiatives.
3. Reflect on the intersection between the AI Justice League's objectives and the ethical principles discussed in Kant's "Groundwork of the Metaphysics of Morals." 4.Discuss and identify key themes or principles from Kant's work that align with the mission and values of the AI Justice League. Where can you find duty, autonomy, universal morals, etc.
4. Instruct students to develop a set of ethical guidelines for AI based on the insights gained from the AI Justice League website and the ethical principles of Kant. You should have 10 guidelines. You can use https://futureoflife.org/open-letter/ai-principles/ as an example.
• Consider how these guidelines can address biases, promote fairness, accountability, transparency, and respect for individual autonomy.
• Think about how these guidelines can be applied in various AI applications and societal contexts.