Homework: Reflect on Readings

Background

As part of this course, you will be given assigned readings. These readings will focus on design, AI, and ethics. You must take the time to reflect on each week’s readings to fully comprehend the material and be fully prepared for the following week’s course.

Learning Objectives

By the end of this activity, you will better grasp the readings assigned for this week. You will have summarized the main concepts and ideas in the papers using your own words. This will enable you to work with the materials more effectively as the course progresses.

Instructions

  • In your notebook, create an entry for each question below, with an answer to the question in your own words.
  • Questions for Crawford & Paglen (2019)
    1. The authors argue that assembling a training dataset is not only a technological project but also a political one. Why?
    2. The authors describe the anatomy of a training dataset consisting of three parts. Can you name each part and explain how they relate?
    3. The authors argue that solving the problem of bias and fairness by making a dataset more granular, balanced, and inclusive is not sufficient. Why is this the case? And what alternative approach do they advocate for?
  • Questions for Gebru et al. (2020)
    1. The paper describes the relationship between datasets and machine learning models. How would you describe this relationship?
    2. The paper argues for the use of datasheets in machine learning. Why do the authors feel this is necessary?
    3. How would you describe what a datasheet is?
    4. The authors say a datasheet should not be automatically created but should always be created manually. Why do they feel this is necessary?
    5. The paper lists several challenges related to the creation and maintenance of datasheets. Select one challenge that you think is interesting and briefly summarise it.

Product

The assigned readings should be summarized in your own words and captured in your notebook as a collection of notes, each containing the summary of one idea.

Follow-up

Generated ideas will be used in next week’s session.