DATA 385 Ethics
Weekly Choice Menu
Logistics
We will be discussing a different theme around Data Science ethics each week. These topics roughly follow the coding/modeling topics that we are learning about in class that week. You are expected to engage with these materials at least 30 minutes each week and share out with the class in Discord (each theme will have a channel). Specifically,
- Pick at least one item from that week’s theme, or find your own source on the theme. If you use your own, include the link and how you found it. If the item you chose is short (e.g. an 8 minute video), engage with two things.
- Post in the theme’s Discord channel about something you notice, connect with, recognize, or wonder about. (60%)
- Reply to at least one classmate’s post in the same channel. (40%)
CautionNYT articles behind a paywall
As a Chico State Student, you can get a Butte County Library card for free here. This gets you access to hundreds of books, audiobooks, movies, newspaper articles and more. If you find a NYT article that you really want to read, see here for how to activate your 72-hour pass for NYT through the Butte County Library.
ImportantItems listed here are not an endorcement of a view
I have tried to include multiple sources and multiple perspectives. This list is not exhaustive and I encourage you to engage critically and dig deeper into topics that interest you.
other stuff
https://datascienceeducation.substack.com/p/when-ai-can-code-what-should-students?utm_source=podcast-email%2Csubstack&publication_id=29031&post_id=220661224&utm_campaign=email-play-on-substack&utm_medium=email&utm_content=play_card_show_logo&r=5r42i&triedRedirect=true
Week 9: Your data and privacy
What companies collect about you, what they infer, who they sell it to, and why the US has no single federal privacy law.
| Type | Item | Notes |
|---|---|---|
| Reading | Target’s pregnancy prediction, three takes: Duhigg (NYT Magazine, 2012), Bourdet (Vice, 2012), and Piatetsky (KDnuggets, 2014) | How Target used shopping data to predict which customers were pregnant before they told anyone. Duhigg is the original report (paywalled, see the library pass above), Bourdet is a short summary, and Piatetsky argues the viral story of a teen outed to her father probably didn’t happen as told. |
| Reading | California Privacy Protection Agency: Personal information and data brokers | What data brokers are, what they collect, and how they infer sensitive details like health, family, and finances from it. |
| Reading | WMD Introduction, or Ch 4, “Propaganda Machine: Online Advertising” | The Introduction defines a weapon of math destruction; Ch 4 shows how online ads target vulnerable people, such as for-profit college recruits. |
| Video | DS in a Box Data Privacy | Why data you share, or data collected about you, can create privacy problems even when it seems harmless. |
| Video | The Guardian: Cambridge Analytica whistleblower | Christopher Wylie explains how Cambridge Analytica harvested Facebook data on millions of people for political targeting. |
| Video | Last Week Tonight: Data Brokers (HBO, 2022) | John Oliver explains how data brokers buy and sell personal data and argues for a federal privacy law (strong language). |
| Podcast | Lawfare Podcast: The Hidden Alliance Between Tech and Government (Byron Tau) | Reporter Byron Tau on how government agencies buy location and other personal data from brokers. |
| Podcast | The Security Ledger, Ep. 206: What might a federal data privacy law mean in the US? | A 2021 discussion of what a US federal privacy law could look like and why one doesn’t exist yet. |
Week 10: Web scraping, can vs. should
Public data is not the same as fair-game data.
| Type | Item | Notes |
|---|---|---|
| Reading | Engadget: Scientists release personal data for 70,000 OkCupid profiles (2016) | Researchers scraped and released 70,000 dating profiles, arguing the data was public. |
| Reading | Apify: Web scraping case law, hiQ v. LinkedIn | The court fight over whether scraping public LinkedIn profiles is legal, written by a scraping company. |
| Reading | Publishers Weekly: Judge rules AI training is fair use in Anthropic copyright case (Nawotka, June 2025) | Authors sued Anthropic for training Claude on books, including copies from pirate sites. The judge ruled that training was fair use but downloading pirated copies was not. The case later settled for $1.5 billion. |
| Reading | WMD Ch 10, “The Targeted Citizen: Civic Life” | How campaigns use harvested data to target voters with tailored political messages. |
| Video | DS in a Box: Web scraping considerations | Legal and ethical questions to ask before scraping a website (9-minute lecture). |
| Podcast | ITRC Weekly Breach Breakdown: Court rules hiQ violated LinkedIn terms by scraping | A short episode on the ruling that hiQ broke LinkedIn’s terms of service by scraping. |
Week 11: AI and LLMs
How LLMs are built and who bears the costs.
| Type | Item | Notes |
|---|---|---|
| Reading | MIT News: Explained, generative AI’s environmental impact (Jan 2025) | How much electricity and water it takes to train and run generative AI models. |
| Reading | WMD Ch 1, “Bomb Parts: What Is a Model?” | O’Neil explains what a model is and argues that models are opinions embedded in math; written before LLMs. |
| Video | Safiya Umoja Noble: Imagining a future free from the algorithms of oppression | How search engines and algorithms reproduce racial and gender bias. |
| Podcast | Karen Hao on Empire of AI (NPR interview, May 2025) | Journalist Karen Hao on the labor, data, and environmental costs behind AI companies like OpenAI. |
| Reading | Anthropic: How AI assistance impacts the formation of coding skills (Jan 2026) | A randomized trial of 52 junior engineers learning a new Python library. The AI group scored about 17% lower on a quiz, and how people used the AI mattered. Published by an AI company. |
| Reading | Infosecurity Magazine: Slopsquatting (Muncaster, April 2025) | A study of 16 code-generating LLMs found about a fifth of the packages they recommended didn’t exist, and attackers can register those names with malware. |
Week 12: Misrepresentation, how charts and statistics mislead
Truncated axes, causal claims from correlations, and hidden uncertainty.
| Type | Item | Notes |
|---|---|---|
| Reading | DS in a Box Lab 9: Conveying the right message through visualisation | A hands-on lab where you redesign misleading visualizations. |
| Reading | WMD Ch 2, “Shell Shocked,” or Ch 3, “Arms Race: Going to College” | Ch 2 shows how finance risk models hid uncertainty before the 2008 crash; Ch 3 shows how college rankings reward gaming the metric. |
| Video | DS in a Box: Misrepresentation | How causal claims, axis choices, and hidden uncertainty make data say more than it does. |
| Video | Alberto Cairo: How charts lie | Data visualization expert Alberto Cairo on the common ways charts mislead readers. |
| Podcast | BBC More or Less: Behind the Stats | A BBC show that checks the numbers behind news stories; any episode works, and it’s on any podcast app. |
| Video | Hans Rosling: The best stats you’ve ever seen (TED, 2006) | Rosling’s moving bubble chart of income and life expectancy across countries and years, and how it breaks myths about the “developing world.” A well-designed chart, to contrast with misleading ones. |
| Reading | Lead Stories: White House steel chart fact check (Feb 2026) | A recent case: a bar chart starting at 80.2 million tons made a 1.2% increase look like a doubling. About 400 words, with a full-scale comparison. |
| Reading | The Challenger O-ring charts, two takes: Friendly (DataVis.ca) and Kosara, “Tufte and the truth about the Challenger” (Eager Eyes, 2011) | Friendly shows the engineers’ original graph beside a redesign, and how leaving out the zero-failure launches hid the cold-temperature pattern. Kosara argues Tufte’s well-known version of the story leaves out important context. |
Week 13: Algorithmic decisions and bias (credit, hiring, justice)
Models that decide who gets a loan, a job, or bail, and who checks them.
| Type | Item | Notes |
|---|---|---|
| Reading | ProPublica: Machine Bias (2016) | An investigation finding that a criminal risk-score tool falsely flagged Black defendants as high risk more often than white defendants. |
| Reading | Ch 6 (hiring), Ch 7 (workplace), Ch 8 (credit), or Ch 9 (insurance) of Weapons of Math Destruction | How models decide who gets hired, how workers are scheduled, who gets credit, and what insurance costs. |
| Video | DS in a Box: Algorithmic bias | How algorithms used in hiring, criminal justice, and other decisions can encode bias. |
| Video | Cathy O’Neil: Weapons of Math Destruction or Kristian Lum: What’s an algorithm got to do with it | O’Neil summarizes her book’s argument; statistician Kristian Lum looks at algorithms used in criminal justice. |
| Podcast | CBC Spark: Weapons of Math Destruction (Cathy O’Neil) | O’Neil discusses how opaque models harm people in work, school, and the justice system. |
Week 14: Facial recognition and surveillance
Error rates that differ by skin tone and gender, and what happens when police rely on them.
| Type | Item | Notes |
|---|---|---|
| Reading | Gender Shades (Algorithmic Justice League) and MIT project page | An audit showing commercial face classifiers were far less accurate for darker-skinned women than for lighter-skinned men. |
| Reading | ABC News: Flock license plate cameras and the nationwide backlash (Zahn, 2026) | How Flock’s license plate cameras spread across police departments, why 50+ localities have canceled or turned them off, and what Flock and its supporters say in response. |
| Reading | TSA facial recognition at airports, three takes: AP via PBS NewsHour (2024), Gizmodo via Sen. Merkley’s site (2024), and Baker (Volokh Conspiracy, 2024) | A short overview of the program and the senators’ push to limit it; a critic’s case for an investigation, with how to opt out; and an opinion piece arguing the privacy and bias worries are overstated. |
| Reading | WMD Ch 5, “Civilian Casualties: Justice in the Age of Big Data,” or the Conclusion | Ch 5 covers predictive policing; the Conclusion covers what O’Neil thinks should be done about WMDs. |
| Video | Joy Buolamwini: How I’m fighting bias in algorithms | Buolamwini on discovering face software couldn’t detect her face, and her work on algorithmic bias. |
| Podcast | The Daily: Wrongfully Accused by an Algorithm | Robert Williams was arrested after a facial recognition system falsely matched him to a crime. |
| Podcast | Fresh Air: Kashmir Hill on Clearview AI (2023) | Reporter Kashmir Hill on Clearview AI, which scraped billions of photos to build a face search tool. |