PHIL150 Exam3

The Four Goods of Work

The goods that just work should provide

• income/economic security

• recognition/status

• meaningful activity

• social contribution (or relationships).

• The argument is that work isn't just instrumentally valuable — these four goods matter for human flourishing and dignity

The Right to Individualized Treatment (Duff and Jorgensen)

The idea that individuals are entitled to be judged on the basis of who they are and what they've done, not on the basis of statistical patterns about groups they belong to. In criminal justice contexts (Duff's wheelhouse), this means punishment should respond to your conduct, not your demographic profile

Opacity

opacity refers to the "black box" problem — the inability of affected parties (or even designers) to understand why a system produced a given output. This matters ethically because you can't contest or explain a decision you can't inspect

Algorithmic Decision-Making

The use of automated systems — typically trained on historical data — to make or inform decisions about individuals (parole, loans, hiring, etc.)

Actuarial Reasoning

Statistical/probabilistic reasoning that predicts individual outcomes based on group-level data.

• The philosophical tension: it may be accurate in aggregate but treats individuals as instances of a category rather than as persons

Algorithmic Bias

When an algorithm systematically produces skewed outcomes for certain groups — often because it was trained on historically biased data, or because it uses proxies correlated with protected characteristics

Human Cognitive Bias

The well-documented tendency of human decision-makers to be influenced by irrelevant factors (race, appearance, fatigue, etc.). This is often invoked in defense of algorithmic systems — humans are biased too, so maybe algorithms are better. The debate is whether this defense actually holds

Why the Four Goods of Work are Matters of Justice (according to G+H)

Gheaus and Herzog argue these goods aren't just nice-to-haves — they're justice claims.

The reasoning goes:

• If work is the primary mechanism through which people access income, recognition, meaningful activity, and social bonds in modern society, then unjust distribution of work (or workless futures) constitutes a genuine injustice, not merely bad luck.

• You'll want to be able to articulate why each good matters and why its denial is a justice issue specifically, not just a welfare one

The Proxy Problem and Why it is a Problem

Algorithms often can't use protected characteristics directly (race, gender, etc.), so they use proxies — zip code, name, purchase history — that are highly correlated with those characteristics.

The problem is two-layered:

(1) it functionally replicates discrimination even without explicit intent

(2) it obscures accountability because the system appears neutral. You should be able to explain why "we didn't use race" doesn't solve the fairness problem

Whether Algorithmic Decision-Making Violates the Right to Individualized Treatment

The argument for violation: algorithms classify you by group membership, not individual facts, which is exactly what Duff/Jorgensen say we're entitled to resist.

The argument against: human decision-makers also use generalizations; the question is whether algorithmic generalizations are categorically different. Strong answers will engage with whether accuracy can justify actuarial reasoning, and whether opacity makes the rights violation worse (since you can't even challenge the basis of the decision).