Joint Activity Potential as a Resource for Socio-Economic Development in the Age of Large Language Models

Firms, regions, and countries with a similar set of resources often demonstrate different performance outcomes, and a substantial part of this variation is related not to a deficit of participants' capabilities but to the non-optimal use of their productive complementarity. The paper introduces the concept of joint activity potential (JAPot) as the maximum surplus achievable through the optimal combination of participants' complementary capabilities and intentions, together with a realisation coefficient φ reflecting the degree to which this potential is actually used. A three-level conceptual model is proposed, linking the structure of complementarity, institutional mechanisms, and the cognitive-communicative infrastructure (as a shared mental model, SMM) that both constrains and enables the realisation of JAPot. It is argued that large language models provide technological means for purposefully working with the SMM, opening up the possibility of governing JAPot as a new type of development resources. On this basis, an updated version of the "technology demand signal" for the AI industry, originally proposed in other works, is presented. Assuming that the required AI solutions can be created, principles are formulated, oriented toward intensive rather than extensive growth: a) through the fuller disclosure of already existing complementarity rather than the accumulation of resources; b) by the expansion of JAPot itself. One of the results discusses the risk of cognitive homogenization, including as a result of the improper use of AI, as a threat to intergenerational cognitive heritage. This paper is considered a vision concept, laying out a conceptual framework and problem statement for a long-term research program, as well as defining landmarks for the development of AI technologies.