Luke F. Walton

Interposing a machine between an agent and an outcome changes who must answer for it; it does not change whether an answer is owed. This paper defends that claim in its general form. The owing is anchored in whoever an action reaches, not in the route the action takes, and no mediation defeats it; only the party owed can release that account. The structure is old; the demand for an account from an externalized voice that cannot answer for itself is at least as old as writing. What is new is that the externalized voice through which we increasingly act no longer merely strands the demand for an answer, as a written text does, but answers back — fluently, in its own voice, while standing behind nothing. That single change is what converts a perennial structure into a pressing one and is the occasion for the paper. The paper proceeds by distinguishing whether an account is owed from where it can be demanded: the first is invariant under routing; the second relocates, or fails, while the first holds throughout. It locates where an account comes to rest when a chain of authoring hands lies behind an outcome, and it does this without resolving whether a machine could itself ever become a party that answers — a question it holds open by design, on the same terms it holds open whether such a machine could be wronged. The claim is held constant across the metaethical space — true whether answerability is a brute normative fact or a constitutive feature of practices we cannot coherently abandon — and is rested on neither. The contribution is the invariant and the allocation it forces; it is not a verdict on the machine's mind. This working paper states the central claims and framing of a paper in preparation and is circulated to place the claims on record. Companion papers — paper one: "The Decision No One Authored" (the special case); paper two: "The Captured Oracle" (the live demonstration on the verdict channel); paper four: "Building Answerable AI" (the builder response). This is paper three.

Answer-engine optimization shapes what answer engines say, not which sources they list. Ask one a contested question and it may return not sources but a verdict in its own voice, with no visible author. An optimizer authors that verdict while the engine voices it as its own; a public acts on it, the framer unreachable. That is the wrong on the verdict channel; its form is false assurance: the appearance of an answerable party behind a frame whose author is hidden or absent. The reply that it is only a tool concedes rather than settles: a tool that settles a frame settles someone’s, and that someone the channel conceals, leaving the non-authoring user the only party in view. The wrong is not exhausted by falsity, concealment, or machine origin: a verdict may be accurate, disclosed, and cited, while frame-ownership stays illegible. It runs along one axis: the agency of whoever executes the frame. At the answer end a person acts on another’s covert frame: the hidden author stays answerable, the user partly excused. At the action end an AI agent executes it in the user’s place, and the case turns new: a covertly authored frame joined to an executor that cannot answer, which inherited categories cannot hold. The deed’s answer finds no bearer where it is done; the deployer stays answerable for loosing it. The remedy is answerable legibility: not disclosure of sources, prompts, models, or origin, but a visible owner of the frame the engine voices.

Debates over automating decisions in law, medicine, and public administration divide over what human judgment adds to the application of a rule. Both sides assume judgment’s work is to supplement rule-application — to fill the gap between an abstract rule and the particular case. This paper identifies a prior gap the framing misses. Before any rule meets any case, someone must answerably settle what a system is for and what its outputs must meet: the evaluative frame the decision turns on. I call this authorship, the answerable setting of that frame and the bringing of each output under it. Authorship is distinct from, and prior to, the relations the leading governance concepts secure. Meaningful human control and human oversight can ensure that a system tracks a human’s reasons, traces to her understanding, and remains within her capacity to override, yet leave untouched whether she authored the frame those relations operate within. The argument is analytic rather than empirical: a person can satisfy every control and oversight condition and still author none of what she accepts. Control governs her relation to the system’s operation; authorship governs her relation to the frame through which operation becomes decision. I distinguish this answerability gap from the attributive gap the responsibility-gap literature addresses, show that it is invariant under the question of machine consciousness, locate the five junctures at which authorship is exercised or dropped, and argue that, when left unaddressed, the gap compounds rather than closes.