Shaxnoza Arifdjanovna Abdullayeva
Leading Engineer, Department of Investments and Capital Construction, Almalyk Mining and Metallurgical Complex; PhD Student
E-mail: sh.abdullaeva@agmk.uz
ORCID: 0009-0009-7722-3492
Scientific supervisor: Z.U.Berdinazarov,
Doctor of Economics, Associate Professor
Abstract. Large industrial investment programs use scheduling, document, enterprise-resource-planning and reporting platforms, yet value is lost when the systems are not connected through common identifiers, governed data and auditable decision processes. This study develops an Artificial Intelligence-Driven Digital Investment Project Management Framework (AI-DIPMF) and a composite performance index (AIPPI) for that integration problem. The study applies design-science research. A structured state-of-the-art review was synthesized with the PEER logic (Point, Evidence, Explanation, Relevance); official AMMC disclosures were used for all quantitative case facts; BPMN-compatible process analysis, ISO 21502, ISO 21508:2026, ISO/IEC 42001, ISO/IEC 23894, ISO/IEC 42005, NIST AI RMF and OECD/JRC composite-indicator guidance informed the artifact and its evaluation protocol. Because no enterprise-wide implementation dataset was available, untraceable before-and-after estimates were not treated as results. The study produces: (i) a governed five-layer architecture connecting enterprise source systems, an integration and semantic-control layer, a project data platform, AI services and human-in-command workflows; (ii) a revised eight-dimension AIPPI with operational definitions, provenance and robustness rules; and (iii) a staged shadow-mode and controlled-pilot evaluation design. Official AMMC records show the relevance of the problem: the first-quarter 2025 portfolio comprised nine major projects with total stated cost of USD 13.5867 billion, while a prior transport-logistics information system was officially reported to improve equipment management by 10%, reduce paper documentation by 70% and save more than UZS 7 billion per year. These figures demonstrate local digitalization potential, not the effect of AI-DIPMF. AI-DIPMF is a testable reference architecture rather than a proven enterprise-wide intervention. Its effectiveness and efficiency should be accepted only after out-of-time model validation and a controlled rollout demonstrate lower forecast error, shorter cycle time, lower reporting effort and positive risk-adjusted economic value without breaching security, reliability or human-oversight thresholds.
Keywords: artificial intelligence; investment project management; design science; composite performance index; megaprojects; mining and metallurgy; digital transformation; AI governance.