An AI-Augmented Decision Architecture for IT Investment: Linking Technical Maturity, Financial Valuation, and Accountable Governance

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Open innovation makes IT investment decisions dependent on knowledge, technologies, and evidence that cross organizational boundaries. Yet these decisions are evaluated through disconnected logics: technical teams assess maturity; executives evaluate cost, strategic value, and flexibility; and artificial intelligence (AI) systems extract evidence or rank alternatives. Their separation creates a translation problem: maturity judgments enter financial models as unexamined assumptions, estimates remain detached from changing technical evidence, and algorithmic recommendations become difficult to contest. This conceptual paper develops the AI-Augmented IT Decision Architecture (AIDA), a closed-loop decision-support architecture connecting an evidence fabric, dynamic maturity assessment, probabilistic valuation, explainable synthesis, and accountable governance. Its design is grounded in a secondary conceptual synthesis of a 39-study core drawn from three previously assembled evidence streams and two post-search updates published in 2026; it is neither a new systematic review nor an empirical validation. AIDA treats evidence bundles, readiness-risk profiles, value-option sets, decision cases, commitment packages, and outcome records as boundary objects coordinating internal and external actors. Twelve propositions specify expected relationships among traceability, uncertainty propagation, option-aware staging, explanation, role separation, feedback, and decision outcomes. The article also contributes a reusable constructed benchmark model: a generative-AI investment case, a fully parameterized 100,000-iteration Monte Carlo implementation, reproducibility files, and a comparative protocol that researchers can reuse, challenge, or extend to evaluate AIDA and alternative approaches. The benchmark demonstrates internal analytical behavior under declared assumptions rather than organizational superiority, while explicit escalation rules distinguish full AIDA from AIDA-Lite. Decision quality is expected to arise not from a single superior method, but from disciplined translation among distributed evidence, economic consequences, and accountable strategic judgment, particularly in open innovation ecosystems where evidence, incentives, and complementary assets span organizational boundaries.

This publication has 52 references indexed in Scilit: