Epistemic Limits and Formal Logic : Learning About the Richness of Traditional Ecological Knowledge Through Artificial Intelligence

Sammanfattning

This thesis investigates the epistemological and technical limits of Artificial Intelligence (AI) when applied to the rich, context-dependent domain of Sámi Traditional Ecological Knowledge (TEK). Motivated by the escalating threats to socio-ecological endurance in Sápmi and the risks of "hermeneutical injustice" posed by statistical AI models, this research employs a constructive design approach. By attempting to encode the narrative practice of the "Sunrise Pasture Assessment" into the formal logic of Social Practice Theory (SPT), the study performs a boundary analysis, using the rigidity of code to highlight precisely what resists digitization. The formalization demonstrates that while SPT can successfully coordinate the high-level syntax of herding, such as roles, resources, and conditional heuristics, it encounters hard structural boundaries when attempting to encapsulate the embodied, relational semantics of the practice. The analysis identifies key epistemic gaps, including ontological blindness to unencoded sensory variables, the erasure of granular herd dynamics, and the inability to process the historical continuity of the Sámi landscape. This thesis reframes digitization bias as possibilities for strategic abstraction. It argues for a variable-resolution approach to AI design, where systems explicitly manage structural landmarks while deferring execution to human epistemic authority. Ultimately, delineating these computational boundaries serves as a mechanism of protection against technological solutionism. It resolves the paradox of participation by proving that while AI may coordinate logistics, the core wisdom of Sápmi resides strictly within the Sámi community and must remain sovereign.