CSE 163

Fairness & Ethics Case Studies

​

Arpan Kapoor�Summer 2026

​

​

​

COMPAS Questions (Fairness)

1. One idea to counter-act systemic racism into models like COMPAS is to explicitly code race into the model, and then make some mechanism to force the model to be fair concerning race. However, in many places, this is illegal to do in the first place! How would you respond to this idea?

​

• Consider the false positives and false negatives in this scenario. What are the risks associated with each, and what would you use to guide your idea of fairness in this case study?
• Loomis v. Wisconsin was a 2017 court case that challenged the state of Wisconsin’s use of COMPAS to sentence a man named Eric Loomis to six years in prison. Read this statement from the Harvard Law Review and reflect on the judges’ decision. Why did the court accept the result from COMPAS? What are the criticisms made of this decision?

“Predicting Criminality” Questions (Fairness)

1. How would you define the construct space, the observed space, and the decision space in this study?
2. The original paper that Wu and Zhang published can be found at this link. What stands out to you about their study? How does reading the Bergstrom and West critique of the paper first affect your perception of it?
3. Wu and Zhang published a response paper to the critiques of their original study. How would you summarize their response? What similarities or differences do you see in the points made in their response and the criticisms that Bergstrom and West wrote?

“Predicting Criminality” Questions (Fairness)

Brandon H. & Brodie Monohan

• How would you define the construct space, the observed space, and the decision space in this study?
• The original paper that Wu and Zhang published can be found at this link. What stands out to you about their study? How does reading the Bergstrom and West critique of the paper first affect your perception of it?

The authors conclude that criminality can to a degree, be determined through specific patterns in facial features. However, s

• Wu and Zhang published a response paper to the critiques of their original study. How would you summarize their response? What similarities or differences do you see in the points made in their response and the criticisms that Bergstrom and West wrote?
• In their response, the authors take a much more neutral approach to the topic. They dispel

Ethics Discussion Questions

• What is the data that is being used in your case study? What assumptions are being made of it?
• What groups are affected by the application in your case study, and in what way? (This can be positive or negative!)
• What privacy concerns might arise from using the data in your case study?
• Have you used the technology in the case study (or one that is similar)? What did you know about it before using it? Did you learn anything new from the case study?

Template Slide

• Make a copy of this slide and replace the title with your/your group members’ names and label it “Fairness” or “Ethics”
• Select a case study from Lesson 9.1
• Discuss and respond to the corresponding discussion questions

Fairness - Joshua, Kateryna

• How would you define the construct space, the observed space, and the decision space in this study?
• Construct space: Facial features indicating criminality
• Observed space: Are they convicted of a crime?
• Decision space: The criminality of the individual
• The original paper that Wu and Zhang published can be found at this link. What stands out to you about their study? How does reading the Bergstrom and West critique of the paper first affect your perception of it?

The authors conclude that criminality can be determined (with meaningful accuracy) through specific patterns in facial features. However,

• Wu and Zhang published a response paper to the critiques of their original study. How would you summarize their response? What similarities or differences do you see in the points made in their response and the criticisms that Bergstrom and West wrote?

It was interesting to see them mentioning cultural differences, when advocating for critics claim about smiling and non-smiling faces.

Rishik, Jyothi - Fairness

​

1. We think that this would result in too much complexity—who gets to decide what is fair? How do we actually implement this?
2. ​
• False positives: If someone is marked as having a higher risk of reoffending even if they don’t necessarily reoffend, it can be harmful to the person trying to get parole. It could also propagate existing prejudices against their specific demographic if it fits into a stereotype.
• False negatives: If someone is not marked as having a higher risk of reoffending, yet they do commit crimes again, it could be harmful to the general society.
3. ​
• They accepted this ruling because the decision was made on publicly available data, and data explicitly given by the defendant. So, they were able to say the sentencing was individualized
• Judges are not allowed to use risk scores alone to determine sentencing and severity, which using COMPAS gets close to doing

​

Fairness: Darren Gao, Aarushi Koneru

• I would argue that while race shouldn’t be a factor in determining certain mechanisms in our models I think the way society is structured almost required. One thing brought up was that race didn’t even need to be inputted into the model as it was able to correlate other factors such as income, address, name, etc. So the model was able to determine race without needing race so I think it would be better to explicitly account race than to account for it later implicitly through other factors.
• False positive could unfairly give harsher sentences as it can risk future harm to others. I would prioritize avoiding unequal false positives across all races as people should not lose freedom due to a biased prediction.
• The court accepted COMPAS because it was only a single factor in sentencing (not referenced as sole evidence). However critics argue that a judge may rely on the score too much despite it’s possible bias.

• Make a copy of this slide and replace the title with your/your group members’ names and label it “Fairness” or “Ethics”
• Select a case study from Lesson 9.1
• Discuss and respond to the corresponding discussion questions

Addie, Kateryna, Jyothi - Ethics

Secret Invasion:

​

• The secret invasion intro used a lot of ai generated art.
• Marvel thought that the ai generated art represented the themes of their story the best, but it also affects artists negatively as it kind of sets a precedent that real artists aren’t needed to make art for these big corporations
• Also it doesn’t say anything about where these models got their art trained on, and the style that they are trying to mimic.
• I haven’t generated any AI art myself, but I did learn about how corporations have been using it in real life.

Tina, Addie, Kelin, Eveie - “Fairness”

1. We would define the construct space to be the contribution of micro-expressions to the identification of criminality, the observed space to be the different scans of faces, and the decision space would be the gauge of criminality.
2. What stood out to me us this study was how strong their claim was, despite the limited evidence and basis the authors made off of their dataset. We find that reading the critique first contributes to the bias of seeing how flawed their initial interpretations were, though it does offer clarity on their actual intentions as opposed to determined conclusions on criminality.
3. In their response, they argued that it was just an exploration of machine learning and its applications, rather than a direct correlation between facial expressions and criminality. We think that one of the main similarities mentioned in both their response and criticisms was on bias and the potential risks of misuse, especially when it comes to machine learning.

​