Home > IT/기술 > Kookmin University Innovates Public R&D Evaluation with Generative AI, Proposing a Framework to Bridge the 'Maturity-Expectation Gap'

Kookmin University Innovates Public R&D Evaluation with Generative AI, Proposing a Framework to Bridge the 'Maturity-Expectation Gap'

Beyond the limitations of public R&D evaluation, strengthening consistency and efficiency with AI.

IT_기술 2026-03-18 IT/기술

Beyond the limitations of public R&D evaluation, strengthening consistency and efficiency with AI. Since the AlphaGo vs. Lee Sedol match in 2016, Artificial Intelligence (AI) has evolved beyond mere technological potential to become a core tool influencing our daily lives and public policy. Recently, a research team led by Professor Kim Do-hyung of Kookmin University's KIBS (KMU International Business School) has garnered attention by announcing a public Research and Development (R&D) evaluation model utilizing Generative AI. The team stated that on March 18, 2026, their research paper on a Generative AI-based public R&D evaluation decision-making framework was published in 'Technovation,' an internationally renowned SSCI-indexed journal in the field of business administration. This research is significant not only for explaining the advancements in AI technology but also for presenting its potential as a practical problem-solving tool in the public sector. Since COVID-19, national and public institutions have faced increasing demands for transparency and efficiency in budget management, and Generative AI demonstrates the potential to support these needs. Public R&D evaluation systems have repeatedly revealed issues in many countries. While public R&D projects are reviewed through annual and phase evaluations to check progress and decide on continuation or direction adjustments, existing evaluation methods have consistently faced criticisms such as a lack of consistency in evaluation criteria, evaluation bias, and efficiency problems in large-scale project evaluations, largely due to their reliance on experts' subjective judgments. Specifically, the subjectivity and lack of consistency in existing evaluation systems have often been cited as factors blurring the direction of R&D projects. To address these limitations, the Kookmin University research team analyzed the potential of Generative AI and proposed a new framework called the 'Maturity-Expectation Gap (MEG).' This framework serves as a tool to quantitatively analyze the difference between the actual technological maturity of public R&D projects and stakeholders' expectation levels, representing a core concept that can elevate the potential application of AI in future public policy. According to Professor Kim's research team, introducing Generative AI into the evaluation system based on the MEG framework is expected to innovatively improve existing inefficient elements. The core of the MEG framework lies in quantitatively measuring the difference between the actual technological maturity of public R&D projects and stakeholders' expectation levels. To achieve this, the research team systematically combined two main methodologies. First, they collected survey data from experts with experience in public R&D evaluation to gain practical insights from the field. Second, they analyzed academic literature related to Generative AI and public policy using machine learning to strengthen the academic foundation. The unique methodological contribution of this study lies in combining these two approaches to compare and analyze technology expectation levels and perceived technological maturity. The analysis revealed significant differences between stakeholder groups regarding their expectations for Generative AI and their perception of its actual technological maturity. It also confirmed a tendency for trust and willingness to adopt AI to decrease as the expectation-maturity gap widens. This suggests that simply introducing technology is not enough; managing stakeholders' expectation levels is a critical factor for successful AI adoption. A particularly noteworthy finding is the ability to diagnose the potential for Generative AI adoption across different evaluation areas, distinguishing between areas where technology application is straightforward and those requiring additional preparation. Such detailed diagnosis can serve as an important guideline for setting priorities and efficiently allocating resources when incrementally introducing AI in the public sector. For instance, in evaluation areas centered on quantitative data analysis, Generative AI can yield immediate effects, whereas areas requiring qualitative judgment or contextual understanding may necessitate further technological development or the establishment of human-AI collaboration systems. The research team emphasizes that such differentiated strategies by area can significantly increase the likelihood of successful AI adoption. What is the Core of the 'Maturity-Expectation Gap (MEG)' Framework? Public R&D in the Korean economy already plays a crucial role in strengthening national competitiveness, with tens of trillions of won invested annually. However, at the same time, public questions about the efficiency of budget execution and the consistency of projects are continuously raised. In this reality, the Generative AI utilization plan proposed by the MEG framework has the potential to go beyond mere