Bridging the maturity-expectation gap: Generative AI in strategic decision-making for public R&D interim review

Public R&D interim reviews face persistent challenges related to scalability, evaluator bias, and inconsistency in multi-stakeholder judgment. Generative AI (Gen AI) has the potential to mitigate these limitations by enhancing efficiency and standardization, yet its deployment also introduces risks such as algorithmic bias and loss of transparency. This calls for a systematic framework to support strategic decision-making and guide responsible adoption. This study introduces the Maturity–Expectation Gap (MEG) framework, which captures the misalignment between stakeholder perceptions of AI maturity and the actual technological state. Existing models such as the Technology Acceptance Model fail to reflect the temporal and institutional dynamics of emerging technologies. To address this, the study combines expert survey data on perceived maturity with machine learning-based literature analysis to compute expectation scores across twenty-four evaluation criteria. Results show that MEG significantly varies across stakeholder groups, and that higher MEG values are associated with lower confidence in Gen AI adoption, highlighting the framework's utility in explaining strategic adoption decisions (RQ1). Furthermore, MEG enables diagnostic classification of evaluation domains, identifying areas of alignment (e.g., Financial Health) and misalignment (e.g., Decision-Making), thereby supporting phased and risk-aware deployment strategies (RQ2). MEG framework offers a structured lens for managing expectation–capability alignment, extending technology adoption theory while supporting strategic decision-making for the responsible integration of Gen AI in public-sector evaluation systems. • This study introduces the Maturity–Expectation Gap (MEG) for Generative AI in R&D reviews. • MEG compares Perceived Maturity (PM) and Technology Expectation (TE) using mixed methods. • High MEG values lower adoption confidence, especially in Decision-Making criteria. • Financial Health and Compliance show low MEG, indicating readiness for phased AI integration. • MEG extends the Technology Acceptance Model (TAM) for multi-stakeholder policy settings.