Introduction

Data is heavily involved in every aspect of AI! The models cannot learn without data. But how do we decide what data to collect? How do we go about actually collecting it? How do we process it once we have data? How do we ethically share data? These are all critical questions! The book we are reading for this module, Data Feminism, takes the viewpoint that data is power. Those who own the data have power. Those who collect the data have power. And it continues throughout the cycle! 

The book is arranged around a series of guiding principles (this is a common theme of our readings this semester!). Since the book is a bit long to get through the whole thing one chapter per class, we are going to let you choose the parts you are most interested in and report back to the class about them. Everyone will read the first chapter and then we will do a bit of an adventure for the next two assignments!  Read on to learn more.

Assignment 1

Read Chapter 1: The Power Chapter. Be prepared to discuss how this chapter relates to AI and Ethics. If you want to learn a bit more about why they are setting things up the way they are, read the introduction as well. 

One group will need to signup to present a summary. Everyone else needs to bring discussion questions.

This chapter is full of great things!  But one specific thing I want you to come prepared to answer, think about something you are working on. It can be your thesis or your class project or anything.  Be ready to answer the following questions from the chapter:

This often means asking uncomfortable questions: who is doing the work of data science (and who is not)? Whose goals are prioritized in data science (and whose are not)? And who benefits from data science (and who is either overlooked or actively harmed)?

The second half of this assignment is focused on case studies. Here, your job is to look through the news and find examples of ways in which data has been collected ethically and responsibly or NOT (more likely to find the NOT in the news).  I have three examples below, which you should read.

• Example of how NOT to collect image data: Largest Dataset Powering AI Images Removed After Discovery of Child Sexual Abuse Material
• Another example (you can just read the abstract here) of poor choices in collection leading to issues: Easily Accessible Text-to-Image Generation Amplifies Demographic Stereotypes at Large Scale
• This one is an example of how to do better! Guidelines on the use of Indigenous data from UNESCO: Leveraging UNESCO Normative Instruments for an Ethical Generative AI Use of Indigenous Data

Post your article (OR your summary for the group that signed up for that) in the Assignment 1 for Data is Power