An interface for a life
What kind of interface between a human being and the digital world could preserve and expand embodied agency?
The question begins with a familiar discontinuity. A life proceeds through relationships, projects, places and obligations, while its digital traces are distributed among applications. A conversation becomes a message, a decision becomes an entry in a calendar, a developing thought becomes several documents. To act, a person repeatedly reconstructs the relations between these fragments. The interface presents one service at a time; the situation continues across them.
ARCHETYPE explores the possibility of an interface organised around that continuing situation. It would help a person work with digital means while remaining able to determine and revise the direction of their participation. This is a philosophical and technological hypothesis. The account that follows describes a possible form of assistance and the demands it would have to meet, rather than a system whose effects have already been established.
Its starting concern is the relation between intellectual assistance and a life that exceeds the interface. How could a system support the continuity of that life without making itself the necessary centre through which everything must be understood?
The digital situation
A digital environment helps determine what can be noticed, remembered and done. An inbox makes some requests prominent. A calendar gives commitments a visible arrangement. Search returns a particular field of answers. A recommendation makes one possibility easier to encounter than another. These forms participate in the situation before a person explicitly asks them for help.
Adding AI does more than accelerate individual operations. A system can also describe what the situation is, connect fragments and propose the next action. Its account may be useful precisely because the person could not have assembled it alone. Yet that account remains a selection. Some evidence is available, some is absent, and some connections are interpretations.
An interface attentive to agency would need to keep those differences accessible. It should be possible to discover what a suggestion rests on, which part is uncertain and what has not been taken into account. An unknown conversation cannot be silently filled in as if it were a known event.
The point is not to require constant inspection of every computational step. It is to prevent a provisional description from becoming an unquestionable world. A person needs ways to correct the account, contribute something the system cannot observe and ask a different question of the same material.
Continuity across tools
Organising an interface around a person's situation would change its basic unit. A message could belong to a relationship and to a continuing undertaking; a document could record a decision whose consequences are still unfolding. The relevant context would include how a process has changed, what remains unresolved and what the person has chosen to undertake.
A conversational window alone does not provide this continuity. Nor does a voice that can launch several applications. The distinctive hypothesis of ARCHETYPE concerns a revisable relation between context, delegated work and the person who remains answerable for its direction. Many different technical implementations might attempt such a relation.
Imagine, as a design question, a journey whose arrangements depend on a meeting. A changed date could affect correspondence, transport and commitments to other people. An interface might make these relations available together and prepare permitted changes. It would also need to recognise the limits of its mandate: the reason for the journey may have changed in a conversation to which it has no access.
Continuity therefore cannot mean a complete model of a person. The model would have to remain incomplete and correctable. A life organised through a single interface could otherwise become easier for the system to administer and harder for its participant to reinterpret.
Delegation that can change
The philosophical distinction developed in AESTHESYS concerns delegating operations while retaining authorship of direction and the criteria for revising it. ARCHETYPE asks how that distinction might become part of an interface.
A delegation would need to express more than a desired result. What may the system do? Under which conditions? When must it return to the person? What would count as completion, and what would require a pause? These boundaries make continuing assistance possible without turning a single instruction into indefinite authority.
The ability to cancel an operation is one part of the relation. A deeper possibility is to revise the undertaking itself. The person may discover that the goal no longer matters, that a constraint was wrongly described, or that the kind of help being offered is narrowing their understanding. An interface should leave room for this change even when the previous arrangement was efficient.
More confirmations do not necessarily produce more participation. A person can become a habitual approver of a direction supplied elsewhere. Equally, bounded operations can proceed without repeated interruption while the person retains a meaningful ability to question the purpose, inspect the result and change the arrangement. The appropriate distribution of attention and authority would have to be established for particular kinds of work, including how it behaves when something fails.
Memory without a fixed identity
Digital memory could support thought by preserving the context that would otherwise have to be reconstructed. In the author's conversations while developing a book, AI was described as making it easier to formulate thoughts, recover connections and continue thinking without the same fear of forgetting. This is a personal report of an experience. It does not establish a general cognitive effect or show what caused the experience.
The observation nevertheless gives a concrete starting point for inquiry. What becomes possible when less effort is spent holding material in readiness? And what may be lost if an external system increasingly decides which material returns to attention?
A memory attentive to agency would have to distinguish an event, a person's account of it and an interpretation proposed by the system. It would need to preserve changes of mind as changes, rather than treat them automatically as errors or inconsistencies to correct. Earlier commitments can remain consequential without becoming permanent definitions of who someone is.
Remembering also involves selection. An interface may make certain episodes easy to retrieve and leave others almost invisible. Its summaries can gradually acquire more authority than the records from which they were made. Correction, access to relevant sources, limits on retention and the ability to remove material would therefore be central design questions. Storing more would not, by itself, make the memory more faithful to a life.
An opening onto the unexpected
An interface could help a person encounter possibilities that do not simply repeat established preferences. A work of music might lead to an unfamiliar artist, a place to an unexpected history, a developing question to someone working in another field. The significance would lie in the possibility of an encounter that the person had not already planned.
Serendipity cannot be reduced to an increased supply of recommendations. A suggestion can broaden a field of attention; a continuous stream can occupy that field so completely that little room remains for anything else. Even an accurate prediction of desire does not tell us whether the form of influence is acceptable.
The difference becomes practical through the relation the interface permits. Can the person understand why something has been proposed? Can they refuse without being treated as an obstacle to engagement? Can they ask for a different kind of possibility, or for no suggestions at all? Can a system disclose when another party has an interest in a recommendation?
ARCHETYPE would have to treat these as conditions of a meaningful invitation. The outcome of an encounter cannot be prescribed. A person may follow a connection, change it, or find that it matters differently from what the system anticipated. The possibility remains theirs to take up, without an obligation to realise the interface's account of their interests.
The body and the limits of a model
A digital system can process descriptions of bodily circumstances and build a model of a context. Such a model does not live the person's situation or undergo the consequences of their action in their place. Information about fatigue is not the fatigue through which a decision is made. A description of a relationship is not direct access to the other person's experience.
This distinction does not make intellectual assistance irrelevant to embodied life. A recollection, a comparison or an unexpected formulation may change what a person can discern and do. Digital work already belongs to a bodily situation: it occupies time, establishes a pace and changes what receives attention.
The interface would therefore need to allow a person to bring their situation back into the account. “I cannot do this now” may concern a condition absent from the available data. “This no longer matters” may revise a criterion that the system has accurately remembered. Neither response should automatically be interpreted as a failure to follow an optimal plan.
Nor should the system claim privileged knowledge of the person's true motives. A pattern in recorded behaviour can justify a question without settling its meaning. The tact required here concerns when and how to make a difference available while leaving its significance open to the person whose life is involved.
Authorship and the possibility of leaving
Authorship of direction includes the ability to change the form of assistance. A person might want execution without interpretation, a counterargument instead of reassurance, a record without a recommendation, or a period without digital mediation. The interface should not require all these relations to take the same conversational form.
This places ethical demands on any attempt to realise ARCHETYPE. Permissions must remain intelligible and revisable. Access to information must not silently become authority to act. Private context requires boundaries concerning its collection, use and disclosure. A person needs ways to object, inspect consequential activity and stop a delegation. These are principles to be worked out and tested; their statement is not a technical or legal guarantee.
Exit is more demanding than a visible off switch. If important memories, commitments and working practices become available only through one system, formal permission to leave may coexist with a high practical cost. Portability, intelligible records and the continuity of essential tasks are therefore part of the philosophical question of agency, as well as questions of engineering and institutional responsibility.
Dependence also needs careful discrimination. An instrument can remain valuable without indicating a failure of human development. The concern is whether a person becomes unable to reconsider the criteria through which the instrument presents their life. Growing operational reliance and decreasing dependence on an external interpreter are compatible possibilities; neither should be assumed in advance.
A hypothesis to be tried
ARCHETYPE proposes a direction for inquiry: an interface that helps sustain context, supports bounded delegation and opens possibilities while preserving the person's access to the question of how to continue. Its coherence as an idea does not demonstrate its feasibility, reliability or effects on agency.
Several tensions would need to be examined in concrete situations. Reducing the effort of oversight may also reduce knowledge of what has happened. A useful memory may become an intrusive interpretation. A well timed suggestion may be difficult to distinguish from influence shaped by another party's interests. The ability to correct a system on paper may prove costly in use.
Research would have to ask what a person can actually understand, refuse and revise, including when assistance is interrupted or produces a misleading account. Fewer interactions, faster completion and reported satisfaction would each tell us something, but none alone would establish an expansion of embodied agency.
“AI in your life. More life beyond AI.” can be read as a question addressed to this hypothesis. What would assistance have to be like for digital intelligence to take a useful place within life while leaving the course of that life open to human participation? ARCHETYPE remains an attempt to give that question a possible interface.