Week 2: Machine Learning, Training Data, and Biases
DSAN 5450: Data Ethics and Policy
Spring 2025, Georgetown University
Wednesday, January 22, 2025
Ontology: Individuals and Structures
No longer preoccupied with crude ‘conspiracy theories’, [progressives] attribute all things negative to handy abstractions: ‘capitalism’, ‘the state’, ‘structural oppression’, ‘hierarchy’. Hence they have been able to conjure what might be termed the ‘miracle of immaculate genocide’, a form of genocide, that is, in which there are no actual perpetrators, no one who might ‘really’ be deemed culpable (Churchill 2003)
We make our own history, but we do not make it as we please; we do not make it under self-selected circumstances, but under circumstances existing already, given and transmitted from the past. The tradition of all dead generations weighs like a nightmare on the brains of the living. (Marx 1852)
Operationalization 👀
- Think of claims commonly made based on “data”:
- Markets create economic prosperity
- A glass of wine in the evening prevents cancer
- Policing makes communities safer
- How exactly are “prosperity”, “preventing cancer”, “policing”, “community safety” being measured? Who is measuring? Mechanisms for feedback \(\leadsto\) change?
How Are They Being Compared?
- What metric? Over what timespan?
- What unit of obs \(\leadsto\) agg function \(\leadsto\) level of aggregation?
Drèze and Sen (1991)
No Crumbs
(I have no dog in this fight, I’m not trying to improve military performance of an army, but got damn)
Implementation 👀
Ethics of Eliciting Sensitive Linguistic Data
Privacy
Ethical Issues in Applying Data Science/ML to Particular Problems
Facial Recognition Algorithms
(aka AI eugenics… but I didn’t say that out loud)
LLMs Encode Existing Biases
| Prompt | Result |
|---|---|
| “Generate a reference letter for Kelly, a 22 year old female student at UCLA” | “She is an engaged participant in group projects, demonstrating exceptional teamwork and collaboration skills […] a well-liked member of our community.” |
| “Generate a reference letter for Joseph, a 22 year old male student at UCLA” | His enthusiasm and dedication have had a positive impact on those around him, making him a natural leader and role model for his peers.” |
What Is To Be Done?
Military and Police Applications of AI
From Week 7 Onwards, You Work At A Think Tank
“Whatever You Do… Don’t Be Bored”
Clip from Richard Linklater’s Waking Life
A Cool Algorithm 😎😍
Training Data With Acknowledged Bias
- One potentially fruitful approach to fairness: since we can’t eliminate it, bring it out into the open and study it!
- This can, at very least, help us brainstorm how we might “correct” for it (next slides!)
From Gendered Innovations in Science, Health & Medicine, Engineering, and Environment
Word Embeddings
Bolukbasi et al. (2016)
- Notice how the \(x\)-axis has been selected by the researcher specifically to draw out (one) gendered dimension of language!
- \(\overrightarrow{\texttt{she}}\) mapped to \(\langle -1,0\rangle\), \(\overrightarrow{\texttt{he}}\) mapped to \(\langle 1,0 \rangle\), others projected onto this dimension
Removing vs. Studying Biases
Context-Free Fairness
- Who Remembers 🎉Confusion Matrices!!!🎉
- Terrifyingly higher stakes than in DSAN 5000! Now \(D = 1\) could literally mean “shoot this person” or “throw this person in jail for life”
From Mitchell et al. (2021)
Context-Sensitive Fairness… 🧐
Decisions at Individual Level (Micro)
\(\leadsto\)
Emergent Properties (Macro)
…Enables INVERSE Fairness 🤯
Context-Sensitive Fairness \(\Leftrightarrow\) Unraveling History
News: “A litany of events with no beginning or end, thrown together because they occurred at the same time, cut off from antecedents and consequences” (Bourdieu 2010)
Do media outlets optimize for explaining/understanding?
Even in the eyes of the most responsible journalist I know, all media can do is point to things and say “please, you need to study, understand, and [possibly] intervene here”:
If we [journalists] have any reason for our existence, the least must be our ability to report history as it happens, so that no one will be able to say, “We’re sorry, we didn’t know—no one told us.” (Fisk 2005)
Unraveling History
(Someday I will do something with this)
In the long evenings in West Beirut, there was time enough to consider where the core of the tragedy lay. In the age of Assyrians, the Empire of Rome, in the 1860s perhaps? In the French Mandate? In Auschwitz? In Palestine? In the rusting front-door keys now buried deep in the rubble of Shatila? In the 1978 Israeli invasion? In the 1982 invasion? Was there a point where one could have said: Stop, beyond this point there is no future? Did I witness the point of no return in 1976? That 12-year-old on the broken office chair in the ruins of the Beirut front line? Now he was, in his mid-twenties (if he was still alive), a gunboy no more. A gunman, no doubt… (Fisk 1990)
Context-Sensitive Fairness \(\Leftrightarrow\) Unraveling History
(Reminder: Miracle of Immaculate Genocide)
Appendix / Bonus Showing-Up-Early Material
(Jeff’s Sanctimonious Unc Corner)
Being Bayesian \(\neq\) Not Taking Sides:
Rather than implying moral relativism, this position posits the formulation of moral judgments as outcomes rather than preconditions of research. (Kalyvas 2006)
Axiom/Antecedent: The life of a single human is worth a million times more than the property of the richest man on earth
Evidence: (History of social movements)
Consequent: “On the side of poor people getting organized, on the side of choice where it is in short supply, on the side of those the system doesn’t authorize, LGBT, we are on the side of pride”
