What the Slogan Does—and Does Not—Claim
A blunt phrase needs a precise operating manual.
The Library / 72 works
Essays, briefings, and index terms. Filter by territory, search by idea, or start with the relationship between two claims.
A blunt phrase needs a precise operating manual.
Why an output can be useful evidence without being a speaker with a stake.
A text can be moving without proving that anything is moved by writing it.
What a sentence resembles, what it does, and what it proves are three different things.
The human feeling of a machine is partly a design decision made by people.
A useful map for locating responsibility in a system that speaks as one thing but is built by many.
A system does not need an inner life for institutions to owe people an explanation.
No interface is immaterial: data, labor, energy, hardware, policy, and capital remain in the room.
A task can be exposed to automation without a worker being replaced, helped, or protected.
Taking uncertainty seriously without converting it into proof or a marketing strategy.
Rights, welfare, liability, and representation are different questions with different consequences.
Public rules can shape AI systems without settling the deepest philosophical questions.
Small changes in language keep questions open without laundering responsibility.
Why one screen can make three different things appear to be the same thing.
What a system can imitate, what a company can promise, and what care requires from people.
Why an output becomes consequential only through a human and institutional chain.
A system can be consistent and still carry choices, priorities, and blind spots.
Why an answer that sounds common can still be manufactured, partial, or strategically useful.
When a system is wrong, the important question is how a person is met afterward.
A friendly interface can make disclosure feel intimate while data practices remain institutional.
After capability, what kinds of judgment do we still expect people and institutions to practice?
First person is a powerful interface affordance, not a measurement of selfhood.
Comforting language can help a user while remaining a product behavior rather than a reciprocal feeling.
An output can look purposive without demonstrating a purpose held by a subject.
There is always a human or institution with a power the interface makes easy to miss.
The smoothest interaction can rest on work that is distant, precarious, hidden, or emotionally costly.
The cloud is a built environment with local costs, not an abstraction that floats above place.
A corpus is not a moral vacuum just because it is large, public, or technically available.
A disclosure that arrives after attachment, reliance, or payment is not doing the work people imagine.
A companion product should not become a substitute for accountability, care, or human contact by default.
An unanswered question is not affirmative evidence for the answer a product wants.
A legal status distributes powers and duties; it does not reveal the inner nature of its holder.
Treat the sentence as a product event before treating it as a report from a hidden subject.
Trustworthy claims need a trail from marketing sentence to operational practice.
A user can click ‘agree’ while still having no meaningful alternative.
A system can feel socially present long before it is socially accountable.
When a model declines a request, an organization has usually made a choice about the interaction.
A system can be accurate enough for one task and still be unworthy of broad reliance.
A citation is a relationship between a claim, a document, and a reader—not a decorative footnote.
A numeric output gains force only when an institution gives it force.
A human contact point is meaningful only if it can actually change the outcome.
Terms can allocate risk without establishing that a person had a meaningful choice.
An account of an automated process needs both a trace of what happened and a truthful way to explain it.
Continuity of product behavior is not evidence of a continuous self.
The option a user sees first can distribute power before anyone makes an explicit choice.
A label can create obligations in a user’s mind before a product has earned them.
A calm, fluent answer can borrow the force of expertise without carrying expertise’s duties.
The identical sentence can be harmless in one setting and dangerous in another.
A demonstration shows possibility; a service needs a record of responsibility.
Measures can sharpen attention, but they can also make important realities invisible.
A responsible system needs someone empowered to stop it before a failure becomes routine.
The capacity to act for reasons or purposes; a word that needs a subject before it can do moral work.
The practice of being answerable for a decision, its grounds, and its repair.
The attribution or presentation of human-like traits, motives, feelings, or social roles.
A contested term often used for subjective experience, awareness, or what it is like to be a being.
A meaningful, informed, and appropriately voluntary agreement; not merely a click.
Information presented in time for a person to understand an automated relationship and make a meaningful choice.
The possible felt or subjective dimension of a life, as distinct from an observable output.
A phrase that may mean real review, nominal supervision, or almost nothing at all.
The designed surface where a system's technical behavior becomes a human relationship.
An entity understood as capable of acting for reasons and bearing responsibility for acts.
An entity whose welfare or interests could matter for its own sake.
A technical component that maps inputs to outputs; not the whole product or institution around it.
A cluster of moral, social, and legal ideas that should not be used as one undifferentiated label.
The selection or generation of an output that fits patterns in data or context.
A request, instruction, context, or interaction frame that shapes a system's output.
What can be lost, gained, damaged, or protected by a decision or relationship.
The linked material, labor, data, policy, and capital conditions that make an AI service possible.
An artifact used toward an end; a category that does not reduce the artifact's power or its maker's responsibility.
A condition in which evidence or theory does not license a settled conclusion.
The condition of being helped, harmed, benefited, or burdened in a morally relevant way.
Language or behavior designed to resemble concern, validation, or understanding without demonstrating reciprocal feeling.
A different route