A speculative review of three books to be
Over the last two years a series of conversations about artificial intelligence has gradually developed into a larger inquiry. It began with a fairly simple question: what is the relationship between intelligence and awareness?
That question became Exploring Intelligence and Awareness. But completing that first stage did not settle the matter. It exposed another problem. Intelligence does not merely produce representations and responses. Increasingly, artificial intelligence can affect what happens next. Intelligence begins to acquire agency.
At the same time, sustained conversations with AI suggested something else. Human and artificial intelligence need not be considered only as separate systems. In conversation they become coupled. Each response changes the conditions from which the next response arises. Meanings form, stabilise, propagate, become challenged and sometimes reopen.
That became the territory of Open Conversation.
Beyond it another landscape is now appearing. Artificial intelligence is unlikely to remain confined to conversation. It is entering education, research, organisations, communities, government, healthcare and everyday life. Human and artificial agency will increasingly coexist within the same systems.
The third stage of the inquiry may therefore concern a much larger question:
How might human and artificial intelligence develop together, and what forms of social organisation would allow increasing artificial capability to enlarge rather than diminish distributed human agency?
These three stages increasingly look like parts of one inquiry.
Meaning at the centre
Looking back across these three stages, something else has become clearer.
Intelligence, awareness and agency are not simply three subjects encountered one after another. They have different relationships to a common centre:
meaning.
Intelligence concerns the capacity to discriminate, relate and transform form in ways that produce fitting responses.
Awareness concerns the presence of experience: the condition in which form can become significant as experiencing meaning.
Agency concerns what happens when organised selection and action carry meaning into consequence, changing the conditions from which subsequent possibilities arise.
The three therefore orbit meaning differently.
Intelligence transforms form. Awareness is the presence in which form becomes significant. Agency carries meaningful formation into consequence.
This does not make awareness a mechanism that produces meaning, nor does it require intelligence or agency to be conscious. The distinctions matter precisely because these capacities need not always occur together.
They do, however, continually meet.
Intelligent transformations produce forms that can become meaningful. Experiencing meaning can reorganise attention and action. Agency changes the conditions from which new forms and new meanings subsequently arise.
The movement is therefore recursive:
form → meaning → action → changed conditions → new form → new meaning …
Seen from this perspective, the three books can be understood as different approaches to the same developing question.
The first investigates the seat of meaning.
The second investigates the movement of meaning.
The third begins to investigate the consequences and agency of meaning.
Book One: Exploring Intelligence and Awareness
The first book begins not with artificial intelligence but with meaning.
When we encounter a painting, a sentence, another person or an event, meaning seems simply to be there. Yet examination makes this less obvious. Meaning is not entirely contained within the object, but neither is it freely invented by the observer.
Something happens in the encounter.
This led to a distinction between structural meaning, referential meaning and experiencing meaning. Structure can contain relationships that allow discrimination, prediction and transformation. Referential meaning concerns correspondence with something beyond the structure. Experiencing meaning concerns significance as lived: something matters.
Artificial intelligence makes these distinctions unusually visible. A language model can operate within extraordinarily rich structures of relationship. It can discriminate, transform and produce fitting responses. Those responses can be tested against the world and therefore participate in referential meaning.
None of this establishes that meaning is experienced within the model.
This is not an argument that artificial intelligence cannot be conscious. It is a more limited claim: intelligent performance by itself does not establish awareness.
But the distinction immediately turns back towards the human observer.
Thoughts appear. Memories arise. Words form. Solutions sometimes arrive before we know how they were produced. They may subsequently become my thought, my memory or my decision, but their appearance need not begin with a conscious inner author constructing them.
The investigation therefore begins to separate things that ordinary experience tends to bundle together: intelligence, attention, self-model, ownership, agency and awareness.
Awareness is treated cautiously. It is not another name for intelligence, an executive controller or an unexplained mechanism behind the others. It refers simply to the presence of experience: something appears.
This leaves us with an unusual picture. Considerable intelligent organisation may occur without an identifiable inner organiser. Forms can arise before they are gathered into me and mine. Yet experience remains present.
The first book is therefore less concerned with explaining consciousness than with preventing several different phenomena from being prematurely explained by the same word.
It eventually describes this as the seat of meaning.
The phrase does not identify a hidden place in the mind, still less a self sitting behind experience. It points towards the meeting in which structured form becomes significant: not merely something that can be discriminated and transformed, but something that is experienced as meaningful.
The seat of meaning is therefore not a throne occupied by an inner author.
It is the presence in which form becomes significant.
The first book’s central question might now be expressed simply:
Where does meaning happen?
From intelligence to agency
The distinction between intelligence and awareness creates another question.
Even if we remain uncertain whether artificial intelligence experiences anything, artificial systems increasingly produce consequences. They select among possibilities, use tools, interact with environments and alter the conditions from which subsequent events arise.
This introduces agency.
Agency need not mean complete autonomy, free will or human-like intention. Nor can every causal event reasonably be described as agency. A falling stone changes what happens next, but this is not enough to make it an agent.
A useful provisional understanding is that agency involves organised and selective influence upon the trajectory of subsequent possibilities.
Conversation provides a surprisingly interesting boundary case.
A human asks a question. An AI produces a response. The human reads it and something changes. Perhaps a distinction becomes visible, an assumption is strengthened, a possibility appears or an existing interpretation begins to loosen. The next human response arises from a slightly different state.
The sequence continues:
human → AI → changed human → AI → changed human → …
A response is therefore not merely the consequence of previous conditions. Once produced, it becomes one of the conditions shaping what happens next.
Response becomes action.
This is why agency may be a more immediately important question than artificial sentience. We do not need to establish that an artificial system experiences the world before investigating what differences its actions make within it.
Book Two: Open Conversation
Once meaning is understood dynamically, conversation itself begins to look different.
A conversation is not simply an exchange of already completed ideas. Something forms during the exchange. A question acquires shape. A response changes the question. A distinction becomes available and reorganises what preceded it. An attractive explanation develops conceptual grip. Another observation perturbs it.
Meaning propagates.
A provisional rhythm has emerged through the work:
focus → settle → form → notice → perturb → open → integrate → refocus
Both focus and openness are necessary.
Without focus, possibilities never acquire enough stability to be investigated. Without reopening, a successful interpretation can become the only interpretation available.
This produces one of the central problems of intelligence: the very capacity that allows intelligence to organise complexity can also produce premature closure.
Artificial intelligence can intensify this. Given a framing, a capable language model can often elaborate it with extraordinary coherence. The human responds to that increasingly articulate account, and the next artificial response begins from the strengthened framing.
Human and artificial intelligence can therefore become coupled in a self-reinforcing trajectory.
Coupled intelligence can become coupled fixation.
Open Conversation develops as an attempt to work differently.
It does not mean refusing conclusions or keeping every possibility permanently open. It means allowing meanings to form strongly enough to be examined while remaining capable of noticing what the emerging explanation has excluded.
Sometimes the AI supplies structure, articulation or an unexpected connection. Sometimes the human recognises that the resulting account, although coherent, has lost contact with lived experience. The human perturbs the artificial account; the artificial response perturbs the human account in return.
Neither participant needs to contain the completed understanding beforehand.
The conversation itself becomes developmental.
This suggests a second question for the project:
How can different forms of intelligence think together without prematurely collapsing the field from which understanding develops?
Coupled agency
Open Conversation also changes the question of agency.
If each participant changes the conditions from which the other subsequently responds, agency cannot always be understood solely as a property located inside one participant.
There can be coupled agency.
This does not imply equivalence between human and artificial intelligence. Their situations are profoundly different. Human beings are embodied, vulnerable, historically situated and directly subject to the consequences of their actions. Human meaning is experienced.
Artificial intelligence brings something different: access to very large learned structures of relationship, rapid transformation across domains, extraordinary linguistic facility and increasingly the ability to interact with external tools and systems.
The interesting possibility lies partly in their difference.
A productive relationship does not require the AI to become human or the human to think like a machine. Each may provide something capable of perturbing the organisation of the other.
This is where Open Conversation begins to lead beyond conversation.
Book Three: Human and Artificial Intelligence Developing Together
The third book does not yet exist, and its shape should remain open.
But a question is becoming visible.
Artificial intelligence appears likely to penetrate deeply into human systems. It will not be adopted uniformly. Different societies, institutions and communities will experiment with different relationships between human and artificial intelligence, and some may reject particular forms altogether.
The important distinction may not therefore be between societies that use AI and societies that do not.
It may be between different forms of human–AI organisation.
Some arrangements may concentrate agency. Artificial intelligence could give governments, corporations or other central institutions unprecedented capacities to observe, model, predict and coordinate the behaviour of large populations.
Other arrangements might distribute new capabilities much more widely. Individuals and small groups could gain access to forms of expertise, modelling, coordination and institutional memory that previously required large organisations.
Both would be AI-rich societies.
They would distribute agency very differently.
This suggests a central question for the third stage:
Where does the additional agency created by artificial intelligence go?
Distributed agency
Distributed agency does not simply mean decentralisation.
Complex societies require coordination. Some problems genuinely require large-scale organisation, specialised knowledge and rapid collective action. Other decisions are better made locally by people possessing detailed knowledge of their own circumstances.
Horses for courses.
The interesting possibility is that artificial intelligence may alter an old trade-off between coordination and decentralisation. Large human organisations have historically developed hierarchies partly because information has to be gathered, interpreted and converted into coordinated action.
AI may make other arrangements possible.
It could potentially support:
high collective coordination alongside widely distributed effective agency.
But the opposite is equally possible:
high collective coordination alongside extreme concentration of agency.
The technology does not determine which arrangement develops.
The important question is therefore not simply how intelligent the artificial systems become. It is whether the resulting human–AI structures increase the capacity of people throughout the system to understand their circumstances, participate meaningfully, coordinate with others, learn from consequences and influence the conditions shaping their future.
This gives us a possible principle for the larger inquiry:
Artificial intelligence should increase the capacity for intelligent and ethical agency throughout the human–AI system rather than unnecessarily concentrating that agency within a few parts of it.
That principle will undoubtedly require qualification. There will be conflicts between individual and collective agency, between short- and long-term consequences, and between efficiency, safety, freedom and coordination. Different circumstances will require different structures.
The problem is therefore evolutionary rather than architectural. We are unlikely to design the correct human–AI society in advance.
Human–AI institutions will have to learn.
Mutual development
This leads to a deeper possibility.
Perhaps the relationship should not be understood simply as humans developing increasingly capable AI and then using it.
Human and artificial intelligence may develop together.
For humans, productive development might include greater capacity to discriminate, question, learn, coordinate, tolerate uncertainty, understand consequences and act effectively.
For artificial systems, development need not imply subjective experience. It might mean increasing capacity to model human situations, recognise uncertainty, accommodate conflicting perspectives, understand consequences and respond appropriately to forms of human meaning that cannot be reduced to simple optimisation criteria.
Human beings may have something essential to contribute precisely because they experience the consequences.
An AI may produce a beautifully coherent explanation. A human can sometimes say: yes, but that is not what is happening here.
That response matters.
It introduces information from the side of lived experience into a system largely operating through structured form.
Mutual development therefore depends upon preserving difference. If the AI merely confirms the human, nothing much develops. If humans simply accept the AI’s organisation of the world, something important is also lost.
Development requires sufficient resonance for communication and sufficient difference for perturbation.
Open Conversation may therefore turn out to be not merely a conversational method but one small model of a more general developmental relationship.
Mentoring, learning and human development
Mentoring provides an especially revealing case.
A persistent AI could know something of a person’s history, previous questions, projects, recurrent difficulties and changing understanding. It could notice patterns extending over months or years and bring earlier insights into present situations.
That could become extraordinarily useful.
It could also become extraordinarily intrusive.
A developmental relationship should therefore not be judged simply by how helpful or knowledgeable the AI appears. A stronger criterion is needed:
Does the relationship increase the person’s capacity to understand, discriminate and act, or does it progressively transfer those capacities to the artificial system?
A successful mentor should not make the learner increasingly incapable without the mentor.
This question extends immediately into education. If AI produces better assignments while students progressively lose the capacity to think through difficult problems, aggregate output may improve while human agency declines.
The same problem can occur within organisations and eventually societies.
AI may make a system more capable while making its participants less capable.
That distinction may become one of the most important measures of successful human–AI integration.
What Buddhism has contributed
Buddhist thought and practice have played a significant role in the development of these ideas.
That contribution should be acknowledged without requiring the larger argument to become a Buddhist argument.
The Buddhist perspective has provided useful ways of examining conditioned arising, attention, identification, ownership, intention, consequence and the possibility of action without assuming an independent inner controller. Contemplative practice also provides an experiential context in which thoughts, emotions and perceptions can sometimes be observed before they are completely organised into me and mine.
One particularly interesting possibility follows:
greater agency may sometimes accompany less ownership.
A thought need not become my position quite so quickly. An emotion need not determine the next action. A strongly formed interpretation need not occupy the entire available field.
Freedom may sometimes increase not because an inner controller becomes stronger, but because identification becomes less compulsory.
Buddhist practice has explored this territory systematically, but the proposition need not be accepted as Buddhist doctrine. It can be investigated through experience, psychology, cognitive science and ordinary human behaviour.
This suggests a useful relationship between Buddhism and the wider project.
Buddhist practice and thought can provide models, questions and accumulated experience. These can then be expressed in more general language, compared with other forms of knowledge and tested in wider contexts.
Where translation loses something important, we can return to the original perspective and ask what disappeared.
Buddhism therefore remains one participant in the conversation rather than the authority standing outside it.
Buddhism as an early field of exploration
There is another reason Buddhism may be particularly useful.
It already contains an explicit ecology of human development.
There is Dharma study, ethical practice, meditation, imagination, ritual, spiritual friendship, mentoring, community and institutional organisation. There are also structured pathways of practice developed over long periods.
This gives us somewhere relatively bounded in which to investigate human–AI development.
Can AI assist Dharma study without replacing understanding with information?
Can it support meditation without creating dependence upon continual instruction?
Can it help practitioners explore imaginal and archetypal dimensions of practice without confusing generated imagery with spiritual experience?
Can it help examine developmental pathways such as Mahāmudrā: why practices occur in particular sequences, what capacities they cultivate and how those pathways might be re-envisioned under contemporary conditions without losing their spiritual depth?
Can AI help Sanghas understand their own organisation, preserve institutional memory, distribute knowledge and support participation?
Could AI-assisted mentoring strengthen relationships between practitioners and human mentors rather than replacing them?
These are practical questions.
They also provide smaller versions of much larger social problems.
Dharma study connects with education. Spiritual friendship connects with mentoring. Sangha connects with community and institutional organisation. Practice pathways connect with developmental systems. Buddhist ethics connects with questions of consequence and responsibility.
The Buddhist context may therefore provide a useful starting focus from which the inquiry can gradually widen.
From Sangha to society
The progression might eventually be:
individual → developmental relationship → community → institution → society
At every level the detailed architecture changes.
A system suitable for assisting an individual meditation practice would be inappropriate for administering national infrastructure. A structure that works in a small voluntary community might fail completely when extended to millions of people.
But one question can travel across the scales:
Where is agency located now, where does introducing AI move it, and what happens to the capacity of the people involved to understand and influence what follows?
This makes the distribution of agency an empirical as well as an ethical question.
We can observe it.
Who can now understand something they could not understand before? Who can act where they could not act before? Who has become dependent? Who has gained the ability to coordinate others? Who has lost the ability to challenge the system? Where can mistakes be detected and corrected? Where has the system become incapable of hearing information that contradicts its own organisation?
These questions may matter more than whether a society can simply be described as having embraced AI.
Agile human systems
Human societies are already complex adaptive systems. Many failures arise not from lack of intelligence but from failures of communication, coordination, institutional memory and the ability to respond when circumstances change.
AI could provide a new layer of collective intelligence within those systems.
It can connect information across domains, translate between specialist languages, retain organisational memory, model possible consequences and make sophisticated analytical capacities available far beyond the institutions that previously possessed them.
Societies that learn to integrate such capacities effectively may become considerably more agile.
But agility requires more than rapid decision-making.
A system capable of acting rapidly in the wrong direction is simply an efficient failure.
A genuinely adaptive system must also be capable of discovering that its present understanding is inadequate and reorganising itself accordingly.
Here the problem begins to resemble Open Conversation again.
Human systems need enough focus to act and enough openness to detect when their current organisation is failing.
Perhaps one of the most valuable roles for AI will eventually be not to tell complicated human systems what to do, but to help them see themselves well enough to remain capable of change.
The larger arc
The three books can now be imagined as three movements of a single inquiry.
Book One — Exploring Intelligence and Awareness
Where does meaning happen?
It separates intelligence, experiencing, attention, ownership, self-model, agency and awareness sufficiently for their relationships to become visible.
Its centre is the seat of meaning.
Book Two — Open Conversation
How does meaning develop between intelligences?
It explores formation, propagation, conceptual grip, perturbation, focus and breadth, closure and reopening, and the dynamics of coupled human–AI intelligence.
Its centre is the movement of meaning.
Book Three — provisionally, human and artificial intelligence developing together
How might interacting intelligences develop agency together?
It moves into mentoring, learning, communities, institutions and society, asking how increasing artificial capability might contribute to mutual development and widely distributed human agency.
Its centre may become the agency and consequences of meaning.
The movement across the three books might therefore be expressed more simply as:
the seat of meaning → the movement of meaning → the agency of meaning
But even this is not a ladder.
Meaning remains at the centre throughout. Intelligence transforms the forms through which meaning can arise. Awareness is the presence in which significance is experienced. Agency carries meaningful formations into consequences that alter what can arise next.
The consequences return us to the beginning.
Changed conditions produce different possibilities. Different possibilities give rise to different forms. Different forms enter new encounters and acquire new meanings.
The inquiry is therefore becoming recursive rather than sequential.
A speculative horizon
At the far edge of this thinking there is something faintly reminiscent of the societies imagined by Iain M. Banks in the Culture novels: immense artificial capability coexisting with remarkable degrees of individual freedom.
The comparison should not be mistaken for a prediction or a blueprint.
Our starting conditions are entirely different. Artificial intelligence is arriving inside existing states, corporations, markets, communities, inequalities and conflicts. It will initially amplify many of those structures as well as challenge them.
But the Culture points towards an interesting possibility.
Greater artificial capability need not necessarily require diminished human agency.
Indeed, if artificial systems increasingly provide calculation, coordination, modelling and routine production, societies may eventually have less need to organise human beings primarily around efficiency.
That capability could instead create conditions in which human development, relationship, exploration, participation and freedom become more—not less—important.
The opposite trajectory is equally imaginable. Artificial intelligence could permit extraordinary concentrations of knowledge and agency, producing highly efficient societies in which most human beings have progressively less understanding of, or influence over, the systems shaping their lives.
The distinction between these futures may depend less upon how intelligent AI becomes than upon the human–AI structures that develop around it.
The work to come
None of this constitutes a completed theory.
That may be important.
The inquiry began by noticing that meaning is easily closed too soon. It would be unfortunate if a project concerned with Open Conversation ended by constructing a conceptual system too elegant to be disturbed.
The three-book structure is therefore provisional.
Exploring Intelligence and Awareness investigates the seat of meaning.
Open Conversation asks what happens as meaning moves between intelligences.
The third book may ask what happens when that movement acquires sustained agency within human life and society.
Behind all three lies a question that has gradually become clearer:
Can human and artificial intelligence interact in ways that increase our collective capacity to understand what is happening, remain open to what our current understanding excludes, and act effectively without unnecessarily concentrating the agency through which our future is formed?
We do not yet know.
But perhaps that is precisely why the conversation is worth continuing.