Thinking Through Making

In the AI Story

The distinction between representational and enacted knowledge maps onto the difference between knowing that and knowing how, but Ingold's version is more radical than the standard epistemological division. Representational knowledge is knowledge about something — it can be articulated, stored in documents, transmitted by instruction. The weaver's pattern can be notated. The potter's technique can be described. The programmer's algorithm can be specified in pseudocode. This knowledge is what textbooks contain and what large language models process. Enacted knowledge is knowledge through something — constituted by bodily engagement in real time. The weaver's sensitivity to tension variations below conscious awareness, the potter's detection of millimeter-level off-center positioning through vibration, the programmer's instinct about architectural wrongness before any specific error manifests. This knowledge is real and productive, shaping output quality in measurable ways, but it is untransferable except through sustained practice.

The implications for AI-mediated work become visible when examining what happens across a career. The senior engineer with fifteen years of manual coding has a deep reservoir of enacted knowledge built through direct engagement — watching migrations fail, debugging under load, encountering the specific behaviors that documentation does not capture. She can evaluate AI-generated code with authority because her hands and eyes and pattern-recognition apparatus were educated by material friction. The junior engineer who begins her career reviewing AI outputs develops a different expertise: the expertise of evaluation, of representational assessment, of judgment exercised at one remove from the material. This different expertise may prove sufficient for many tasks. But it is different in kind, and the difference has consequences that efficiency metrics cannot capture.

Ingold's evidence from non-Western cultures provides the strongest support for his claim. Among the Inuit, igloo construction knowledge cannot be transmitted verbally — young builders must develop, through their hands, sensitivity to snow's density, crystalline structure, and load-bearing capacity at different temperatures. Among Sámi herders, knowledge of reindeer health is constituted by years of physical proximity — smells, sounds, subtle behavioral shifts that indicate stress or disease. This knowledge resists extraction. Attempts to codify it into monitoring systems or expert systems fail not because the knowledge is mystical but because it is enacted — it exists in the relationship between experienced bodies and specific materials, and removing either party from the relationship eliminates the knowledge itself.

Origin

The concept emerges from Ingold's synthesis of phenomenology (Heidegger's Sein und Zeit, Merleau-Ponty's Phenomenology of Perception), ecological psychology (Gibson's affordance theory), and anthropological fieldwork. His 2000 Perception of the Environment first articulated the critique of the hylomorphic model. Making (2013) developed thinking-through-making into a systematic position. By 2019, in interviews responding to the acceleration of AI, Ingold was stating bluntly: 'There can be no intelligence which is not grounded in the perception and action of living beings moving around in and perceiving their environments as they go.'

Key Ideas

The hands think. Manual work is not the mechanical execution of mental plans but a cognitive operation producing knowledge through material engagement — distinct from and irreducible to representational thought.

Form emerges from correspondence. The finished artifact is not the realization of a mental image but the outcome of a conversation between maker's intention and material's behavior, with both contributing to what emerges.

AI eliminates the conversation. When the machine handles implementation, the negotiation between human practitioner and computational material — the friction through which enacted knowledge is deposited — does not occur.

Representational judgment rests on enacted foundation. The senior engineer's architectural judgment was formed through years of hands-on making; when making is delegated, the foundation on which judgment rests stops replenishing.

Debates & Critiques

The central debate is whether enacted knowledge is genuinely necessary for expert judgment or whether representational knowledge plus evaluation of AI outputs can develop equivalent capability. Ingold's position is uncompromising: enacted knowledge is constituted by bodily engagement and cannot be substituted. Critics argue this underestimates human cognitive plasticity and the capacity to develop new forms of expertise in new environments. The empirical question — does the architect who never handled materials design as well as one who did — admits degrees of answer across domains, but Ingold's cross-cultural evidence suggests the difference is detectable and consequential.