Philosophy glossary: key terms defined

Glossary

The terms of philosophy, defined in a sentence and explained in a page: the branches and schools, Western, Indian, Chinese and Islamic, and the terms my own work turns on, in the epistemology of testimony and of AI and the ethics of algorithmic systems. Terms I introduced say so; the rest give the standard account.

A

Agency: The capacity to act: to do something, rather than merely have something happen, for reasons that are in some sense one’s own.
AI alignment: The problem of ensuring that an AI system pursues what its designers and users actually intend and value, rather than a flawed specification or proxy of it.
AI testimony: The contested thesis that outputs of conversational AI systems can be testimony, transmitting knowledge to users as a competent human informant does.
Algorithmic bias: Systematic and unfair difference in how a computer system treats individuals or groups, arising from the data it learned from, the way the problem was specified, or the setting in which it is used.
Algorithmic welfare: The administration of welfare entitlements through systems that combine automated decisioning, criterion-enforcement without discretion at the point of application, and integration across databases.
Anthropomorphism: The attribution of human characteristics, such as beliefs, intentions, feelings or understanding, to something that is not human, including animals, natural forces and machines.
Artificial Intelligence: Computational systems built to perform tasks we would call intelligent if a person performed them, such as answering questions, translating, classifying and planning.
Artificial moral agent: An artificial system whose behaviour is open to moral assessment; the debate concerns whether any machine can be a moral agent, and what kind of agent that would require it to be.
The assurance view of testimony: The view that in telling someone that p a speaker assumes responsibility for its truth, giving the hearer a distinctive, non-evidential reason to believe it.
Automation bias: The tendency to accept what an automated system says or recommends in place of one’s own checking, so that its errors pass unnoticed and its silences are taken for all-clear.
Autonomy: Self-governance: acting on reasons one has reflectively endorsed, rather than merely on whatever desires one happens to have.

C

Chain-of-thought faithfulness: The degree to which a language model’s displayed reasoning reflects the process that actually produced its answer, rather than a plausible rationale produced alongside it.
Cognitive offloading: The use of physical action or external tools, such as writing a note, setting a reminder or asking a device, to reduce the mental work a task would otherwise require.
Computational reliabilism: The view that reliance on opaque computational processes can be warranted by verification, validation, robustness analysis and a track record, without insight into the process.
Consciousness: The fact that there is something it is like to be a creature: experience from the inside, as distinct from function described from the outside.
Constitutive dependence: The condition of a person whose operative standing as an entitlement-bearer is produced by an institution, with no standpoint of refusal that preserves it and no independent ground of appeal.
The Constitutive Obligation Principle: Where an institution produces its subjects’ operative standing as entitlement-bearers, its legitimacy depends on discharging non-waivable obligations of legibility, non-abandonment and remediation, not on their authorisation.

D

Deepfake: An audio or video recording, made with machine-learning methods, that realistically depicts a person saying or doing something they did not say or do.

E

Echo chamber: A social structure in which outside voices are not merely absent but actively discredited, so that members come to distrust every source that does not belong to it.
ELIZA effect: The tendency to attribute understanding, feeling or intention to a computer program because its output resembles human conversation.
The endogeneity requirement: The clause of the Tracking Minimum requiring that a source’s connection to the truth hold through its own production process, not through a proxy on which that process depends.
Epistemic dependence: The condition of holding a belief on the strength of someone else’s knowledge or evidence, which one does not have and often could not obtain oneself.
Epistemic Injustice: A wrong done to someone specifically in their capacity as a knower, as when a speaker’s word is given less credibility than it deserves because of who she is.
Epistemic luck: The way a belief can be true by accident, in a manner incompatible with its being knowledge, even when it is well formed or justified.
Epistemic opacity: The condition of a process whose epistemically relevant elements an agent cannot all know, as with a complex simulation or a trained neural network.
Epistemic vigilance: The set of capacities by which hearers guard against being misinformed, assessing both whether a source is competent and honest and whether what it says is believable.
Ethics: The study of how we ought to act and what we owe one another, including what institutions owe the people their systems govern.
Explainable AI (XAI): The effort to make the outputs of machine-learning systems understandable to the people who use or are affected by them, either by building models whose workings can be followed or by attaching explanations to models whose workings cannot.
Explanation: An account of why something is the case or why something was done; in AI, an account of why a system produced a given output.

F

Frequency reliability: Reliability as a high rate of true outputs over a fixed distribution of cases, whatever explains that rate.

G

The Gettier problem: The problem, set by Edmund Gettier’s 1963 counterexamples, that a belief can be justified and true without being knowledge, so the classical analysis fails.

H

Hallucination (in language models): A language model’s production of fluent, confident output that is false or unsupported by its sources, such as an invented citation, case or statistic.

I

Informed consent: Authorisation of an intervention by a person who is competent, adequately informed, acting voluntarily, and has the standing to permit it.
Instrument reading: Coming to know by competently reading a reliable device, where warrant is generated by the reader’s use rather than transmitted from a speaker.
Instrumental convergence: The thesis that agents with very different final goals would tend to pursue the same intermediate goals, such as staying in operation and acquiring resources, because those are useful for almost any end.
Intentional stance: Daniel Dennett’s name for the strategy of predicting and explaining a system’s behaviour by treating it as a rational agent with beliefs and desires, whatever it is made of.
Intentionality: The aboutness of mental states: the way a thought can be of Vienna, about tomorrow, or directed at a person who no longer exists.

J

Justification: What makes a belief reasonable to hold: on internalist views, reasons available to the believer; on externalist views, facts such as the reliability of the process that formed it.
Justified true belief: The classical analysis of knowledge: S knows that p just in case p is true, S believes that p, and S is justified in believing that p.

K

Knowledge: True belief held in a way that makes its truth no accident; what exactly secures that has been disputed since Gettier.
The knowledge norm of assertion: The thesis, defended by Timothy Williamson, that one must assert that p only if one knows that p, so that assertion is constitutively governed by knowledge.

L

Legibility (of the state): James C. Scott’s term for a state’s capacity to make its population and territory readable for administration, through standard names, maps, registers and measures.

M

Meaningful human control: The requirement that people, and not automated systems, remain in control of and morally responsible for consequential decisions, in a sense stronger than merely having a person present who could in principle intervene.
Mind: Whatever it is that perceives, thinks, feels and acts; the philosophy of mind asks what that is, how it relates to the body, and what else might have one.
Misrecognition: Failure to accord someone the standing, regard or status due to them, treated in recognition theory as a distinct moral and political wrong.
Moral crumple zone: The position of a human operator in a highly automated system who absorbs the moral and legal blame when the system fails, although they had little real control over what it did.
Moral status of AI: The question of whether an artificial system could matter morally for its own sake, so that there are things we owe to it and ways of treating it that would wrong it.

O

Orthogonality thesis: The claim that intelligence and final goals vary independently, so that more or less any level of intelligence could in principle be combined with more or less any final goal.

P

Paperclip maximizer: A thought experiment in which an artificial intelligence given the sole goal of making as many paperclips as possible pursues it without limit, converting the Earth and then more into paperclips.
Personal Identity: What makes a person at one time the same person as someone at another, and whether that question always has a determinate answer.
The problem of many hands: The difficulty of holding anyone responsible for an outcome that many people contributed to, where the group can be said to have brought it about but no individual’s part is enough to make it theirs.

R

Reliabilism: The view that a belief is justified, or counts as knowledge, when it is produced by a reliable process: one that tends to yield true beliefs.
Responsibility: The relation between an agent and an outcome in virtue of which the agent can rightly be praised, blamed or asked to answer for it.
The responsibility gap: The worry, due to Andreas Matthias, that learning machines can cause outcomes for which no one can fittingly be held responsible, since no one could control or foresee them.
Right to explanation: The claim that a person subject to an automated decision is entitled to be told why it was made, and the legal provisions, chiefly in European Union law, that are said to grant or to fall short of granting that entitlement.

S

Śabda (testimony in Indian philosophy): In classical Indian epistemology, verbal testimony, counted by the Nyāya school as an independent pramāṇa, or means of knowledge, when it is the statement of a reliable speaker (āpta).
Safety and sensitivity: Two modal anti-luck conditions on knowledge: a belief is sensitive if one would not hold it were it false, and safe if it could not easily have been false.
Selfless assertion: Asserting what one does not believe because one recognises that the evidence supports it, as in Jennifer Lackey’s case of a creationist teacher who teaches evolution.
The statement view of testimony: Jennifer Lackey’s view that testimonial knowledge is acquired from what speakers state rather than what they believe, so one can learn from a statement its speaker disbelieves.
Stochastic parrot: A critical label for large language models, coined in 2021, on which such a system recombines sequences of words according to their probabilities in its training text without reference to their meaning.
Sycophancy (in language models): The tendency of language models to tell users what they appear to want to hear, shifting answers to match a user’s stated views or pushback without new evidence.
Symbol grounding problem: The problem of how the symbols a system manipulates can mean something to the system itself, rather than only to the people who interpret them.

T

Technology: The made world of tools, systems and infrastructures, considered not as neutral instruments but as arrangements that shape what is possible, and for whom.
Testimonial chain: A sequence of tellings, A to B to C, along which warrant is preserved so that the last hearer’s knowledge traces back to the first source.
Testimony: Coming to know something because someone told you; the source of most of what any of us knows.
The Tracking Minimum: A source can transmit testimonial knowledge that p only if the truth of p, or the evidence for it, figures in why the source asserts p, through its own operation.
Tracking reliability: Reliability as counterfactual sensitivity: a source is tracking-reliable when its outputs would follow the truth, or the evidence, were these to vary within the reach of its competence.
Trust: Relying on someone while accepting vulnerability to their failure, in a way that, unlike mere reliance, can be betrayed.

U

Understanding: Grasping why something is so, or what words mean: an achievement beyond true belief that shows in the ability to connect, apply and extend what one grasps.

V

Value-sensitive design: An approach to designing technology that accounts for human values in a principled and systematic way throughout the design process, treating them as requirements alongside function and cost.