The Interface Between Minds

A registrar finishes a night shift and hands over a ward of twenty patients to the day team. She has perhaps ten minutes. In her head sits a rich model: which patient is stable, which one worried her at 3 a.m. without quite triggering an alarm, which family needs a careful conversation, which blood result is pending and would change everything. What she can transfer is far thinner: a few sentences per patient, some numbers, a gesture toward the man in bed seven: “keep an eye on him”. The day team will reconstruct what they can and get on with it. Mostly this works. Sometimes the thing she could not quite encode is the thing that mattered.

That gap, between what one mind holds and what another mind reconstructs, is the subject of this book. More exactly, the book asks what happens to the gap when intelligence enters it:

What form does communication take when intelligence becomes part of the medium through which meaning moves?

Everything else follows from taking the gap seriously.

Three joints

Look at what happened in the handover. There was a sender with an intent: beliefs, worries, priorities. There was a representation: spoken sentences, a printed list, a glance. And there was a receiver doing reconstruction: building a working model of the ward from those thin materials plus everything they already knew.

INTENT  →  REPRESENTATION  →  RECONSTRUCTION
(sender)      (what travels)      (receiver)

INTENT  ≠  REPRESENTATION  ≠  RECONSTRUCTION

The representation is not the meaning. It is something the sender produced, which the receiver uses, with effort and background knowledge, to rebuild something close enough to act on. We say “the message says” as though meaning travelled inside the words like water in a pipe. It does not. When the rebuild succeeds we call it understanding, and forget how much work the receiver did.

Two further separations sit underneath that one, and they will keep returning.

expression  ≠  intent
word  ≠  concept

What someone says is never all they meant, and a word is not the idea it points at. The sender’s intent is richer than any expression of it, and the same word can point at different ideas in different heads. Both gaps are part of why the rebuild is work.

Here, concept is a working name for a distinction or relation an expression may carry, not a claim that minds store wordless tokens.

When it fails (the missed allergy, the misread tone, the instruction followed literally but wrongly), the failure can sit at any of the three joints. The sender may have misjudged what the receiver already knew. The representation may have been unable to carry the load. The receiver may have reconstructed confidently and incorrectly. These are different failures with different remedies, and a great deal of confusion about language, and about AI language, comes from collapsing them into one.

One question can be asked at every joint, and it will keep returning in this book: what had to survive? Not everything. The registrar cannot transmit her whole mind and should not try. What had to survive was whatever the day team needed in order to act well: the worry about bed seven, the pending result. That is a question about purpose, and it cannot be answered by looking at the words alone.

What the past shows, and what it doesn’t

Human beings coordinated in speech long before anyone wrote anything down. Speech carries far more than vocabulary: rhythm, emphasis, pause, gesture, gaze, shared surroundings. Writing made a different bargain. It let a message travel without its speaker, at the cost of stripping away most of that surrounding channel. What remained had to be rebuilt by a reader who was not there.

Walter Ong’s Orality and Literacy (1982) remains the most careful account of how deep that bargain runs: how writing, and later print, reorganised not just storage but habits of thought: lists, definitions, the idea of a fixed text that can be checked word against word. It is worth reading, and worth reading sceptically. Later scholars have shown that oral cultures were never simple, that literate cultures never left orality behind, and that the line between them was entangled with schooling, class and power. The mechanism survives the qualification: changing the representation changes what is easy to think, check and coordinate, without determining any of it.

Printing extended the logic. Elizabeth Eisenstein argued in The Printing Press as an Agent of Change (1979–1980) that standardisation, preservation and cumulative correction (identical maps, comparable editions, errata that could actually propagate) changed the conditions for the Renaissance, the Reformation and the sciences. Adrian Johns and others answered that early print was often corrupt, that careful manuscripts were common, and that fixity had to be built, not assumed. Eisenstein replied that duplication still changed the conditions of accumulation even if every copy was imperfect. The dispute is more useful than either side alone. It shows a pattern this book will meet repeatedly: a new communication technology changes the constraints on what can be preserved, compared and corrected, while the consequences depend on institutions, incentives and use. Communication is one explanatory factor among several, never the single engine of history.

What the past licenses is narrower and more interesting than a grand story. Each shift (speech to script, script to print, print to electronic media) changed three things together: how much of intent could be encoded, how far the encoding could travel without its author, and what a receiver needed to know already in order to rebuild it. Coordination scaled because encodings became cheaper, more comparable and more correctable, not because minds merged.

Intelligence at the ends, then in the middle

Intelligence has already arrived at the ends of the channel. Writing tools suggest replies, finish sentences and adjust tone for the sender. Reading tools summarise and answer questions for the receiver. Hancock, Naaman and Levy (2020) gave this a name, AI-mediated communication: intelligent agents that modify, augment or generate messages on behalf of communicators. In terms of the three joints, the sender’s tools help produce the representation, and the receiver’s tools help with the rebuild.

But in every one of these cases the channel between the two stays passive. Speech, writing, print and broadcast all carried a fixed encoding from one end to the other. Skill sat at the endpoints: the eloquent speaker, the careful author, the practised reader. The channel did no thinking.

The step this book takes is to put the intelligence in the middle. An AI system can operate inside the channel: take a representation produced for one context and transform it into one suited to another mind, another task, another moment. Summarising is the trivial case. The general case is mediation: deciding what form information should take, what deserves attention, how it should reach this person now, and eventually what bounded exchanges can happen between systems without a human reading every intermediate message.

Note what this claim is and is not. It is not that AI “understands” in the human sense, or that generated text is a person speaking. It is the narrower engineering claim that transformations between representations (prose into diagram, an hour of video into five checkable points, forty papers into the three that bear on your current problem) can now be performed by machinery, not only by authors and editors at one end or by effortful readers at the other. That machinery is fallible and lossy and needs measurement. But the structural fact stands: for the first time the channel can do work.

An example fixes the idea. A colleague sends you a twenty-page incident report on a database outage: timeline, logs, five candidate causes, a disputed conclusion. Until recently you had two options: read it, or search within it. A mediating channel offers a third. It can read the report against your current purpose (you are deciding whether to approve the proposed fix) and produce the representation that serves it. A causal diagram of the five hypotheses with the evidence for each. A table of what each hypothesis predicts against what the logs show. A two-paragraph brief with every claim linked to its page.

None of these is the report. Each is a derived view, compiled from some working interpretation of the report, useful exactly to the extent that the interpretation is faithful and the form fits your task. If the diagram drops the one exception that kills hypothesis three, it is not an enhancement. It is a failure. So there are really three things in play: the source, which is the material as received; the representation put in front of you; and what you actually notice and do with it. Each step between them can lose something, and every step needs a way back to the one before. A view without provenance is a rumour with good typography.

Loss, and gain

Return to the ward and suppose a system has read the overnight notes and charts and prepares the handover. It might surface what the registrar’s tired memory left out: the blood result flagged at 03:10, a downward trend in one patient’s observations. That is a gain. It might equally compress her hesitant “keep an eye on him” into the tidy word “stable”, and delete the one doubt that mattered. That is a loss. The same kind of transformation produced both. What separates them is what survived.

So every transformation in this book gets the same two-sided audit:

SOURCE → TRANSFORMATION → REPRESENTATION → RECONSTRUCTION → SURVIVED / FAILED

The failures are familiar once named: a missing fact, a lost qualification, a reversed relation, an unsupported addition, confidence that moved, a trail back to the source that vanished, an intent that changed. Later chapters add the failures that appear when a system starts to adapt to a person or act for one: personalisation that goes too far, inference nobody authorised, action nobody permitted. The gains are just as real: structure made visible, context supplied, effort saved, access widened to someone who could not otherwise use the material. A book that only counted losses would conclude that the safest channel is the dullest one. A book that only counted gains would be selling something. The aim is to keep both in view, against the same question: what had to survive?

Four questions, in order

If the channel can think, four questions open in sequence, and they organise the four Parts of the book.

Representation. What form should carry this meaning? A fixed page need not stay fixed prose. Images, diagrams, tables, timelines, related evidence and fact-checks can be compiled from the same source, if each transformation keeps what matters. The first Part builds toward a Semantic Browser, a renderer that treats documents as sources and checks every transformation for what survived.

Attention. What deserves to reach me at all? Your subscriptions hold forty hours of conference video, and the papers, podcasts and threads pile up past any capacity to read them. Herbert Simon’s old point, that a wealth of information creates a poverty of attention, now applies at personal scale. The second Part follows information from an overwhelming stream down to the small amount that should enter a day, and asks when.

The person. How should communication reach me now? The same content lands differently at a desk and on a train, in five minutes and in fifty, for implementation and for curiosity. The third Part asks what an interface may learn about a person and use, and what it must never infer. Sometimes the right output is less, or nothing.

Delegation. What happens when humans stop writing and reading every intermediate message? A system that knows your work can sometimes predict a bounded choice, draft a reply or negotiate a meeting, and two such systems can exchange thousands of messages no human reads. The fourth Part asks where authority comes from when a system acts for someone, and ends at the far edge the book is named for: meaning moving between people through machine representations, with humans inspecting consequences and not every step.

Each question is forced by the failure of the one before it. Representation without attention drowns the reader in beautiful views of everything. Attention without personal context cannot tell relevant from new, or urgent from merely interesting. Personalisation without delegation leaves every exchange waiting on human keystrokes. And delegation without everything earlier, without provenance, without a way to check what survived, without policies a person can inspect, would manufacture unaccountable action. The order is the argument.

How the book speaks

A short word on method, so the reader knows what kind of statement they are reading. Three kinds appear. Established background is what prior research supports independently of this book, stated with its populations and limits. Argument is what this book proposes: a mechanism, a distinction, a policy that prior work motivates without settling. Proposal is a design for a test that has not been run. This book reports no experiments of its own, and where a study is described it is a design, not a result.

One more orientation. The book describes a single imagined system, a layer that sits between people and AI systems as they communicate, and it calls that layer Language, like the book. Each part adds to the same layer: how it represents and checks what it passes on, how it chooses what to bring forward, how it adapts to a person, and what it may do or say for one. The names met along the way, the Semantic Browser, the Radar, the personal layer, are stages and views of that one layer and not separate products. Whether such a layer is worth building is the question the book ends on.

Preference goes to peer-reviewed work, proceedings, datasets and official documentation; preprints are labelled as such; vendor claims are attributed, not repeated as findings. Correlation is not upgraded to causation, preference to performance, simulation to evidence about people, or benchmark scores to general ability. Where research disagrees, the disagreement stays visible. And one word is used loosely for now: in these early chapters “meaning” means roughly what a sender intended to communicate and what a receiver reconstructed. Nothing here claims a machine has captured some complete inner meaning. Later the book says precisely what a system can represent and how to check it, claim by claim.

Where this leaves us

Language is an interface for rebuilding meaning in another mind, across the joints of intent, representation and reconstruction. Earlier media changed the constraints on that interface without determining history. Intelligence has begun to arrive at the endpoints, and the step this book examines is its arrival in the channel itself, where it can transform, select, adapt and eventually act. That makes the question what survived? a permanent companion, and it makes the answer two-sided, because a transformation can improve a message as readily as it can damage one.

Before asking what a channel with intelligence can do, it is worth asking how much the receiver was always doing. If meaning is rebuilt and not delivered, then what a message means depends on the person receiving it as much as on the words. What, exactly, does a receiver rebuild?

References

• Ong, W.J. (1982). Orality and Literacy: The Technologizing of the Word. London: Methuen. (Routledge reprints, including the 30th anniversary edition, 2012.)
• Eisenstein, E.L. (1979–1980). The Printing Press as an Agent of Change: Communications and Cultural Transformations in Early-Modern Europe. Cambridge: Cambridge University Press. (Paperback edition 1982.)
• Johns, A. (1998). The Nature of the Book: Print and Knowledge in the Making. Chicago: University of Chicago Press.
• Eisenstein, E.L. (2002). “An Unacknowledged Revolution Revisited”. American Historical Review, 107(1), 87–105.
• Hancock, J.T., Naaman, M. & Levy, K. (2020). AI-Mediated Communication: Definition, Research Agenda, and Ethical Considerations. Journal of Computer-Mediated Communication, 25(1), 89–100. DOI 10.1093/jcmc/zmz022.
• Simon, H.A. (1971). Designing Organizations for an Information-Rich World. In M. Greenberger (ed.), Computers, Communications, and the Public Interest. Baltimore: Johns Hopkins Press, pp. 37–72.