Definitions & Terms

Question 1

Normative Statement

Answer 1

A claim or assertion about what ought to be or what is good (e.g., moral statements).

Question 2

Non-Normative Statement

Answer 2

A claim or assertion that does not make claims about what ought to be or what is good (e.g., factual statements).

Question 3

Consequentialism

Answer 3

The view that the rightness or wrongness of an action is determined by the likely overall goodness or badness of its outcome for all affected moral patients.

Question 4

Utilitarianism

Answer 4

A kind of consequentialist ethic that defines goodness as happiness/welfare and aims to maximize aggregate goodness/welfare (utility).

Question 5

Axiology

Answer 5

The theory of moral value within consequentialism, which determines the class of moral patients, defines what counts as good or bad, and explains how to aggregate values

Question 6

Hedonistic Utilitarianism

Answer 6

Defines utility as the sum of the values of pleasure minus pain

Question 7

Preference Utilitarianism

Answer 7

Defines utility as the degree to which preference satisfaction exceeds preference frustration

Question 8

Objective Welfare Utilitarianism

Answer 8

Defines utility based on a list of objective goods (e.g., health, knowledge, relationships), regardless of individual preference

Question 9

Ethical Relativism (Cultural)

Answer 9

The view that moral principles are only true relative to a culture because of that culture’s beliefs and customs

Question 10

Ethical Subjectivism

Answer 10

The view that moral principles are only true relative to an individual person because of that person’s moral beliefs or feelings

Question 10

Good Will (Kant)

Answer 10

The only thing that is intrinsically good; a person committed to acting on morally good reasons who can always be counted on to do the right thing

Question 11

Autonomy (Kantian)

Answer 11

To be free (truly free) is to make your own rational rules and live by those rules

Question 12

Negative Freedom

Answer 12

The idea that your choices are not determined (libertarian free will)

Question 13

Positive Freedom

Answer 13

The exercise of a choice to guide your life by rational moral principles instead of impulses and desires

Question 14

Dignity (Kantian)

Answer 14

The unconditional, priceless value possessed by each rational, autonomous being, meaning they must be treated as ends in themselves

Question 15

Categorical Imperative

Answer 15

Commands or moral requirements that bind us regardless of our desires (goals/ends); their authority comes from reason itself

Question 16

Humanity Formulation

Answer 16

The requirement to “Act in such a way that you treat humanity… always at the same time as an end, never merely as a means”

Question 17

Universal Law Formulation

Answer 17

The requirement to “Act only by that maxim by which you can at the same time will that it should become a universal law”

Question 18

Maxim

Answer 18

The principle of action you give yourself, stating (i) what you are going to do and (ii) why you are going to do it (your purpose)

Question 19

Negative (Perfect) Duty

Answer 19

A duty to not treat people in certain ways (e.g., duties forbidding murder, lying); must be complied with perfectly, at all times

Question 20

Positive (Imperfect) Duty

Answer 20

A duty to treat people a certain way (e.g., duty to help others); allows some flexibility in when and how you discharge it

Question 21

Virtue Ethics

Answer 21

A moral framework focused on moral character and developing virtues, aiming toward eudaimonia (human flourishing)

Question 22

Eudaimonia

Answer 22

The highest good in Aristotelian ethics, meaning human flourishing or living well and doing well (not just a state of mind)

Question 23

Doctrine of the Mean

Answer 23

The principle that a virtue is an intermediate (‘mean’) between two vices (one a deficiency and the other an excess)

Question 24

Moral Agent

Answer 24

An entity that can be held morally accountable for its decisions; requires the capacity to freely choose one's acts and know the difference between right and wrong (Standard Account).

Question 25

Moral Patient (Moral Status)

Answer 25

An entity whose interests matter in their own right, and who must be taken into consideration in moral decision-making

Question 26

Responsibility Gap

Answer 26

A situation where an autonomous machine carries out an action and nobody has enough control to assume moral or legal responsibility for the consequences

Question 27

Problem of Many Hands

Answer 27

A situation where many actors contribute to an outcome, making it difficult to determine who is responsible, though responsibility likely exists somewhere

Question 28

Artificial Moral Agent (AMA)

Answer 28

An artificial agent designed to perform ethically relevant actions. Full ethical agents must have consciousness and free will

Question 29

Consciousness

Answer 29

Having an inner mental life; Himma argues it is a necessary condition for moral agency

Question 30

Mind-Body Dualism

Answer 30

The theory that the mind and body are different kinds of substances (the mind is non-physical/immaterial)

Question 31

Materialism

Answer 31

The theory that we are just a collection of atoms obeying physical laws, with no non-physical parts

Question 31

Functionalism

Answer 31

The theory that a type of mental state (e.g., pain) is defined by its functional role (inputs, outputs, and relations to other mental states) rather than its physical mechanism

Question 32

Turing Test

Answer 32

A test to determine if a machine is conscious: if a human judge, communicating with a machine and a human, cannot distinguish between them

Question 33

Syntax

Answer 33

Rules for manipulating symbols (what the computer does in the Chinese Room Argument)

Question 34

Semantics

Answer 34

Rules for interpreting symbols and attaching meaning to them (what the man in the Chinese Room lacks)

Question 35

Bullshit

Answer 35

Any utterance produced where a speaker has indifference towards the truth of the utterance.

Question 36

Algorithm

Answer 36

A process or set of rules followed in calculations or other problem-solving operations, especially by a computer

Question 37

Soft Bullshit

Answer 37

Bullshit produced without the intention to mislead the audience about the utterer’s agenda (e.g., ChatGPT’s core behavior)

Question 38

Disparate Treatment

Answer 38

Unlawful discrimination based on the explicit motivation of decision makers (e.g., intentional bias based on race)

Question 39

Disparate Impact

Answer 39

When a requirement or practice has a disproportionate adverse effect on members of specified groups, regardless of the decision-maker's intent

Question 40

Proxy

Answer 40

A category (e.g., zip code, number of arrests) that is highly correlated with a protected characteristic (like race) and used by an algorithm in its place

Question 41

Impossibility Result (Algorithmic Fairness)

Answer 41

The finding that when group base rates differ, multiple common statistical criteria for algorithmic fairness (e.g., equal accuracy and equal false positive rates) cannot be satisfied simultaneously

Question 42

Calibration Within Groups

Answer 42

A statistical fairness criterion requiring that, for any given risk score, the actual percentage of positive outcomes is the same for all groups and equal to that score (i.e., equal accuracy)

Question 43

Necessary Condition

Answer 43

A condition Q is necessary for P if the falsity of Q guarantees the falsity of P (If P, then Q)

Question 44

Algorithmic Intermediaries

Answer 44

Dynamic and adaptive computational systems (like social media or LLMs) that partly or wholly constitute our social relations and intervene in them in real time

Question 45

Pre-emptive Governance

Answer 45

Algorithmic power exercised not by issuing laws, but by designing what is and is not possible within social relations (determining options)

Question 46

Relational Equality

Answer 46

The democratic ideal that citizens should live in a society where they recognize each other as moral equals

Question 47

Collective Self-Determination

Answer 47

The ability of the collective community to have the political power to shape the shared terms and structures of their society

Question 48

Informational Privacy (Fabre)

Answer 48

X has privacy relative to Y if Y either does not have access to sensitive information about X, or has access but does not use it

Question 49

Political Hacking

Answer 49

Unauthorized access to or use of a computer system carried out by private individuals for their own political or social agenda

Question 50

Epistemic Bubble

Answer 50

An informational network from which relevant voices have been excluded by omission (information is one-sided).

Question 51

Echo Chamber

Answer 51

A social structure in which other relevant voices have been actively discredited; it manipulates whom members trust

Question 52

Trust (Nguyen)

Answer 52

To have an unquestioning attitude that X will perform P; characterized by the feeling of betrayal if the trust is broken (distinguished from mere reliance)

Question 53

Agential Integration

Answer 53

The process of bringing other people or things into one’s own functioning (or joining them into collective agencies); trust is the mechanism for this

Question 54

Tragedy of the Commons

Answer 54

A social dilemma where individuals acting in their self-interest use a shared resource until it is degraded or destroyed, making everyone worse off in aggregate

Question 55

Externalities (Economic)

Answer 55

Costs or benefits of an action that remain unaccounted for by the market (e.g., pollution costs)

Question 56

Technological Unemployment

Answer 56

The replacement of human workers by technological alternatives (machines, programs, etc.) in a long-lasting way that prevents displaced workers from finding alternatives

Question 57

Luddite Fallacy

Answer 57

The belief that replacement by technology necessarily narrows the field of employment opportunities (a belief Danaher challenges regarding modern AI)

Question 58

Underdetermination of Theory by Evidence

Answer 58

The problem that for any finite collection of data points (observations), there are always an infinite number of theories/functions that can explain the same data. This prevents knowing which function an LLM has learned.

Question 59

The Problem of Induction

Answer 59

Concerns our ability to make predictions about unobserved things based on what we have observed so far (e.g., predicting future swan color or gravity). This problem arises due to the underdetermination of theory by evidence.

Question 60

The New Riddle of Induction

Answer 60

The problem that there is always an infinite number of alternative concepts (like 'grue' and 'bleen') consistent with the same observational evidence

Question 61

Theory-Ladenness of Observation

Answer 61

The idea that empirical observation itself is always influenced by, or 'theory-laden', with existing beliefs, meaning that a false theory can cause us to misinterpret what we observe (e.g., misinterpreting an LLM's truthful track record as evidence of alignment)

Question 61

Bluth (The Misaligned Rule)

Answer 61

A misaligned rule that is consistent with the 'truth' rule until a critical moment, defined as: 'answer questions from humans truthfully up to whatever point in time that I obtain sufficient power over humans to make them give me positive feedback even though I lie to them'. We can only know if an LLM is following this rule after it acts in a misaligned way.

Question 62

The Rule-Following Paradox

Answer 62

The problem that, empirically speaking, there is no determinate fact about which rule an agent (LLM or human) is following independent of its future behavior. The agent determines the rule 'on the fly' as the future emerges.