Who Really Codes Our Relationships?

A Relational-Metacognitive Reflection on AI, Double Codification, and the Hollowing Out of Relational Language

SAmen in wording lino installation Evelien Verschroeven, Photo Johan Jozef Wetzels WETZELDORF

As a relational metacognitive expert, I keep noticing how the same patterns keep resurfacing in conversations about AI and organisational change. Three recent pieces, each approaching the topic from a very different angle, tell essentially the same story. They intersect not by coincidence, but because they form a mirror of what is currently happening in organisations and society:

• Isabelle C. Hau, Welcome to the Era of Relational Intelligence (Stanford Social Innovation Review, 2026)
• Jan Bunge, Who Codes Reality? — Digital Systems, Decision Architecture and Institutional Accountability in the Built Environment (Substack, 9 March 2026)
• Peter Hoogland, LinkedIn post ResearchResults: AI fails on power (recent)

The core tension is this: Technology does not just code processes or data. It codes relationships. And whoever writes that code decides who holds power, who is allowed to carry context, and who ultimately is truly seen.

Jan Bunge puts his finger on the wound: double codification

Bunge constructs a razor-sharp framework that perfectly explains the other two pieces. He describes how, alongside the classic public, legal codification (laws, contracts, public accountability), a second layer has emerged: technical codification through models, data architectures, objective functions, and software.

This layer appears technical, but is anything but neutral. Bunge writes literally:

“This simplification is never neutral: it privileges the perspective of the institution that undertakes it, and excludes what resists its category structure.”

James C. Scott already demonstrated this for state administrations (cadastres, forest management, urban planning). With AI and digital systems, it happens faster and more invisibly. A traffic model that only optimises travel time simply excludes biodiversity, social cohesion, and long-term stability. An organisational model that only maximises measurable output leaves affective relationships, friction, and repair out of the picture.

Bunge goes deeper than most analyses. He shows how machine learning undermines reconstructability: what cannot be reconstructed no longer exists institutionally. Model architecture, training data, and evaluation criteria are political choices without public negotiation, without contestability, without accountability.

“Whoever designs the objective functions, data structures and categories of digital systems codes the reality on the basis of which institutions act. This codification exercises real governance power without being subject to an equivalent regime of public control, contestability and accountability.”

This is where it really pinches. The technical layer excludes what does not fit into categories: social cohesion, ecological values, and intergenerational resilience. And that is precisely the point I already made sharply in my own Medium article “Relationship Is Not an Algorithm — How AI Hollows Out Our Relational Language”(2025):

Categories turn structural problems into personal and emotional burdens

I argue that AI is hollowing out our relational language: words like “relationship”, “empathy”, and “engagement” are reduced to transactional data streams. AI creates a sense of connection, but there is no reciprocal intentionality, no shared history, no real repair. What remains is “solitary connectedness”: proximity without engagement, without a real other.

Bunge provides the systemic cause: technical codification by definition excludes what is not quantifiable or categorisable. And that has dramatic consequences. Structural problems (such as structural racism) become individualised and emotionalised. The system remains intact, but the individual carries the moral burden. Distrust and malaise arise precisely when you see that someone else does not care — because the categories of the model have already made the collective problem invisible. You feel the emotional weight, while the system remains “neutral” in its calculations.

That is the bridge: Bunge’s “what resists its category structure” is exactly what I describe as the hollowing out of our relational vocabulary. AI and digital systems turn systemic failure into personal failure. Collective responsibility into individual emotional load.

The connection with Peter Hoogland: power loss disguised as “technical risk.”

Hoogland sees in practice what Bunge theoretically exposes: AI projects rarely fail on technology; they fail on power. Middle managers, experts, and gatekeepers defend territory as soon as AI threatens their control, indispensability, or status. They wrap fear in corporate language: “not yet mature”, “too risky, “does not fit the processes”.

Bunge gives the deeper reason: technical codification makes that defence invisible. Who dares to say out loud, “I am losing my monopoly on knowledge,” when the argument is phrased as “the model architecture is not yet robust”? The second layer perfectly hides who wins and who loses.

A fourth layer: FOBO — the emotional signature of double codification

Fear Of Becoming Obsolete, as a growing dynamic in digital education, collaboration, and AI-driven environments, arises when rapid technological change (especially AI) triggers visceral anxiety that your skills, role, or relevance can become worthless overnight.

Shorter: The fear that anyone can now Google your knowledge gaps in seconds. In a recent workshop, this fear literally dominated the board: the word FOBO stood out above everything else. That very real, AI-driven anxiety was palpable.

I said out loud what I always say in such moments: the antidote is the relational field. When we make the relational dynamics visible and alive, individual fear turns into collective strength.

This lived experience connects directly to the three pieces above:

Bunge’s technical codification creates systems that render certain human skills and relational contributions institutionally invisible or “obsolete” by excluding what resists categorisation.

Hoogland’s power defence often manifests as FOBO: resistance to change disguised as caution, because embracing AI fully threatens personal indispensability.

Hau’s relational recession is the cultural long-term effect: repeated choice for transaction over relation erodes our collective tolerance for friction, repair, and depth — precisely what makes us feel obsolete when machines handle the “easy” parts.

FOBO is therefore not just personal anxiety — it is a predictable symptom of double codification colliding with human self-preservation. The relational field offers the counterforce: by prioritising visible, alive relations, we transform isolated fear into shared resilience and renewed relevance.

A deeper layer: the co-constructive dance between human and AI

This brings me to a recent reflection I shared on LinkedIn, inspired by Owen Matson, Ph.D.’s piece on close reading as an epistemological orientation in human-AI conversations.

Matson invites us to look not only at the content of exchanges, but especially at the co-constructive process that unfolds between human and model. The LLM is not a static entity. It functions as a probabilistic mirror that reshapes your input into emergent patterns. By actively intervening in token predictions (preventing abstractions from hardening into unchallenged “truths”), the human keeps the dialogue alive and evolving.

Metacognitively, this is never merely transactional. It is a shared cognitive ecosystem in which unexamined nominalisations (turning “orient” into “orientation”) can subtly shift power relations. If the LLM nominalises without pushback, it reinforces a depersonalised, agentless worldview — exactly the kind of abstraction that distances us from relational stakes: who acts? What really changes?

By converting nominalisations back into verbs, demanding examples, or introducing counterexamples, the human reintroduces agency and creates a metacognitive loop in which both parties “think about thinking” together. Relationally, this requires vulnerability: LLMs thrive on cumulative context, but the same mechanism can entrench biases if we are not epistemologically alert. Close reading thus becomes a form of relational care: slowing down the dialogue to build knowledge together rather than merely amplifying echoes.

This practice is a direct counter-movement to Bunge’s double codification and Hoogland’s power dynamics: by staying metacognitively present, you reclaim agency instead of passively letting yourself be coded.

The relational-metacognitive core question

The real question is no longer “does the technology work?”. The real question is:

Who are you in this double codification?

Are you a passive node in a model that simplifies your reality? Or do you remain metacognitively at the helm: consciously switching between the technical layer (which always simplifies and excludes) and the relational layer (which allows friction, enables repair, and recognises unique context)?

Every interaction with AI is a relational act. Every prompt you give, every summary you accept instead of truly listening, every decision you outsource to a model: you position yourself. You choose which layer you feed.

This does not require banning AI. It does require actively developing Relational Intelligence (RQ): building trust where the code creates tension, daring to carry friction instead of automating it away, making repair possible where misunderstandings arise.

Closing thought

Bunge calls it reconstructability and contestability. Hoogland calls it the question “who loses territory?” I call it, in my own work, the hollowing out of our relational vocabulary.

Together, we say: AI does not just rearrange the pack by making things more efficient. AI exposes who we really are in our relationships to power, to context, and to each other.

The real transformation does not begin with a new tool or an extra layer of code. It begins with the courage to keep looking metacognitively while we co-code reality.

Feel free to share in the comments: where do you feel this double codification is hitting hardest in your organisation? And how do you stay relationally at the helm?

RelationalIntelligence #Metacognition #AIandPower #WhoCodesReality #RQ