Skip to main content Link Search Menu Expand Document (external link)

Homework: Reflect on Readings

Background

You are assigned readings throughout this course. The readings cover subjects including design, AI and ethics. Each week, you are asked to reflect on your readings, to help you get the most out of the material, and be properly prepared for next week’s course. 

Learning Objectives

After completing this activity you will have gained a deeper understanding of this week’s readings, and will have summarized the core ideas in the papers in your own words, so that you can work with them in the course going forward.

Instructions

  • In your personal notebook, create an entry for each of the questions below, with an answer to the question in your own words.
  • Questions for Burrell (2016)
    1. The author outlines three distinct types of opacity: (1) corporate secrecy; (2) technical literacy; and (3) complexity of scale. Can you briefly describe each type, and the main solution proposed in the paper for each type?
    2. Two brief case studies are used to show how “the machine thinks”. What is the fundamental difference between how machines reason, and how people reason, according to the author?
    3. In the final section the author discusses the problem of interpretability, and ultimately arrives at the conclusion that efforts to solve it by offering explanations of how inputs are related to outputs are doomed to failure. What are some of the reasons the author gives for this?
  • Questions for Mitchell et al. (2019)
    1. The authors propose model cards as a way to solve a specific problem related to pre-trained models and its impacts. Can you summarise the problem and how model cards address it?
    2. The authors argue for a specific type of evaluation, which they call “disaggregated evaluation” – can you explain in your own words what that means?
    3. The authors distinguish between classification systems and score-based systems, and describe an approach to evaluation metrics for each type: a confusion matrix, and pinned area-under-the-curve (pinned AUC). In your own words, can you explain each type of metric?

Product

The result of this activity is a collection of notes, each with a summary of one idea in the assigned readings, described in your own words, captured in your personal notebook.

Follow-up

You will use the ideas generated in this way in next week’s session’s activities.