I study how artificial intelligence reshapes scientific discovery, technological innovation, industrial competition, and public policy. My work combines responsible AI, innovation economics, patent analytics, scientometrics, network science, and sentence-embedding based text analysis.

I am currently a Responsible AI Researcher at the Responsible AI Center, AI Future Lab, KT Corporation, and a Lecturer in Business Analytics at the Graduate School of Technology and Innovation Management, UNIST. I received my Ph.D. in Technology Management, Economics, and Policy from Seoul National University in 2024.

Interactive semantic map of AI topics across policy reports, academic papers, and patents. The project builds a four-level topic hierarchy:

• L0: analytical domain - Policy, Science, Technology
• L1: broad thematic family
• L2: intermediate semantic cluster
• L3: fine-grained AI key phrase

The visualization supports reference-space exploration, Korea activation, and relative topic-gap analysis across five-year periods.

• Live site: AI Topic Space
• Repository: deep1003/AI_Topic_Space.github.io
• HuggingFace: https://huggingface.co/datasets/deep1003/ai-topic-space-korea

• Moral Orientation and Calibration: ICML 2026 Pluralistic Alignment Workshop
  Camera-ready workshop paper on Judge LLM evaluation, pluralistic moral alignment, and diagnostic decomposition of human-LLM disagreement into moral orientation and moral calibration.
  Reproducibility release: deep1003/moral-orientation-calibration
  Announcement: LinkedIn post

• KidnapRAG: A Black-Box Attack for Hijacking Reasoning in Agentic Retrieval-Augmented Generation Systems
  ACL ARR 2026 May Submission, preferred venue: EMNLP. This collaboration with Prof. Buru Chang's research group at Korea University studies black-box poisoning attacks against Agentic RAG systems. The paper proposes KidnapRAG, a sequential attack that uses three role-specific poisoned documents - Bait, Chain-Link, and Mal-Ins - to attract initial retrieval, redirect query reformulation, and inject attacker-controlled evidence into the reasoning chain.
  Keywords: RAG attack, Agentic RAG, AI security, black-box poisoning, retrieval-augmented generation. Public manuscript link pending.

• Responsible AI, trustworthy AI, AI safety, and AI governance
• Generative AI, large language models, RAG systems, and value alignment
• Agentic RAG security, black-box attacks, and retrieval poisoning
• Patent analytics, scientometrics, and AI innovation measurement
• Economic complexity, technological specialization, and national competitiveness
• Network science, knowledge-space modeling, and semantic embedding methods
• Digital economy, AI regulation, and science-technology-policy interfaces

• Responsible AI governance and value alignment: governance frameworks for high-risk AI domains and culturally contextualized AI value alignment.
• Moral alignment for generative AI: developing methods that connect pluralistic human values, multi-agent evaluation, and responsible deployment of generative AI systems.
• Agentic RAG security: studying black-box poisoning attacks and defense implications for retrieval-augmented generation systems, including the KidnapRAG project under ACL ARR review.
• Generative AI and LLM patent analytics: studying dominant design emergence, technological distance, and cross-national competition using Sentence-BERT, PatentSBERTa, clustering, and ensemble modeling.
• RAG attack and defense mechanisms: developing benchmark datasets, defensive methods, and tooling for retrieval-augmented generation systems.
• AI innovation and national competitiveness: analyzing AI specialization, industrial co-evolution, and policy implications using patents, scientific papers, and economic complexity indicators.

• Ph.D., Technology Management, Economics, and Policy, Seoul National University, 2024
  Dissertation: Exploring complementary innovation of AI technology from a multidimensional knowledge network approach
  Link: HAL thesis record

• M.Sc., Technology Management, Economics, and Policy, Seoul National University, 2019
  Thesis: Heterophily Effects on Industrial Innovation Leadership Changes in Autonomous Vehicle Industry

• Chun, Y., Hur, J., & Hwang, J. (2024). AI Technology Specialization and National Competitiveness. PLOS ONE.
• Chun, Y., & Hwang, J. (2024). The Nexus of Artificial Intelligence and Green Innovation: A Cross-Density Analysis at the Country Level. Journal of the Knowledge Economy.
• Hur, J., Chun, Y., Hwang, J., & Kim, K. (2024). The moderating role of design innovation in the relationship between technology complexity and firm performance. Technology Analysis & Strategic Management.

• Chun, Y., Park, Y., Yoon, J., & Lee, S. (2025). Method, Server and Computer Program for Dynamically Adjusting Value Information.
  A value-alignment method for generative AI systems that combines multi-agent debate and reinforcement-learning-based reward modeling.

I teach business analytics and applied AI methods for graduate students, covering:

• Python and R for statistical modeling, machine learning, and visualization
• Deep learning and machine learning foundations
• AI innovation, generative AI, and responsible AI
• Research design and thesis supervision

• Programming: Python, SQL, R, LaTeX
• AI and NLP: BERT, GPT, Sentence-BERT, PatentSBERTa, RAG systems
• Analytics: network analysis, embeddings, clustering, UMAP, scientometrics, patent analytics
• Statistics and visualization: STATA, SPSS, SAS, Tableau, Power BI, Gephi, NetMiner

• Excellent Patent Award, KT Corporation R&D Center, Invention Competition 2025
• Best PhD Paper Award, Korea Society for Innovation, Management, and Economics, 2024
• Reviewer for Humanities and Social Sciences Communications, PLOS ONE, Journal of the Knowledge Economy, and European Research on Management and Business Economics
• ORCID: 0000-0002-6877-6230

• Email: deep1003@snu.ac.kr
• Google Scholar: Youngsam Chun
• LinkedIn: Youngsam Chun