Selected work

Representative Papers

AI-Driven Radiology Report Generation for Traumatic Brain Injuries.

Jan 30, 2025Journal of imaging informatics in medicine

To address diagnostic delays caused by delayed interpretation of cranial trauma imaging in emergency settings, this study proposes an end-to-end AI system integrating AC-BiFPN and Transformer architectures for multi-scale feature extraction from CT/MRI scans and automatic generation of natural-language radiology reports. The framework uniquely co-optimizes lesion detection accuracy and report semantic coherence in the cranial trauma domain: AC-BiFPN enhances multi-scale lesion localization, while the Transformer captures long-range semantic dependencies to produce structured, clinically interpretable reports. Evaluated on the RSNA Intracranial Hemorrhage dataset, the model achieves significantly higher diagnostic accuracy and report quality compared to conventional CNN-based approaches. This work advances emergency department efficiency and provides an interpretable, deployable solution for clinical decision support and medical education.

3 citationsRead paper

The tragedy of the cognitive commons: collective intelligence beyond AI-induced knowledge collapse

Jul 14, 2026

This study proposes the concept of a “tragedy of the cognitive commons” to elucidate how artificial intelligence agents may trigger a self-reinforcing degradation of humanity’s shared knowledge base—termed “knowledge collapse.” When AI substitutes for individual contributions without adequately reconstructing reliable public signals, systems risk converging to a low-knowledge equilibrium. Integrating dynamic modeling, commons theory, and empirical evidence—including a 25% decline in knowledge sharing on platforms like Stack Overflow—the paper systematically examines the complementarity between general and contextual knowledge and the associated learning externalities. The authors offer five structural critiques of prevailing models, delineate their limitations, and advance a research agenda centered on effort elasticity and governance of knowledge commons, thereby providing a theoretical foundation for improving mechanisms of human knowledge aggregation.

0 citationsRead paper

Biodiversity Media Narratives and Stock Market Performance: Evidence from Europe

Jun 18, 2026

This study investigates whether and how media coverage related to biodiversity influences stock market performance in Europe. Drawing on GDELT global knowledge graph data from France, Germany, Italy, and Spain between 2015 and 2025, the authors construct a media-based biodiversity risk index and employ a panel Granger causality test alongside an augmented inverse probability weighting (AIPW) event study design, complemented by quantile robustness checks. The analysis provides the first empirical evidence that European equity valuations are significantly driven by biodiversity-related media narratives: negative sentiment depresses stock prices, with peak effects emerging three to ten months after the initial shock. Notably, the positive impact of low-risk events substantially outweighs the negative impact of high-risk events, revealing a pronounced asymmetry. These findings remain robust after controlling for market volatility and policy uncertainty.

0 citationsRead paper

Humanity in the Age of AI: Reassessing 2025's Existential-Risk Narratives

Dec 01, 2025

This paper critically examines the empirical foundations of the “superintelligence-induced human extinction within a decade” claim promoted by two bestselling 2025 works—*AI 2027* and *If Anyone Builds It, Everyone Dies*. Method: Drawing on critical technology studies, empirical analysis of generative AI trends (2023–2025), and structural economic observations regarding compute and capital concentration—and integrating Whittaker’s algorithmic accountability and Zuboff’s surveillance capitalism frameworks—the study systematically evaluates the “intelligence explosion → superintelligence → goal misalignment” causal chain. Contribution/Results: Findings indicate that contemporary large language models remain narrow, statistical tools lacking autonomous evolution or runaway capability. The “existential risk” narrative functions less as a technical forecast than as an ideological apparatus legitimizing surveillance capitalism expansion and compute monopolization, deeply entangled with the 2025 AI financial speculation bubble. This work is the first to identify AI doomsday discourse as a capital-driven ideological construct rather than a falsifiable technological prediction.

0 citationsRead paper

Can we cite Wikipedia? What if Wikipedia was more reliable than its detractors ?

Sep 02, 2025

This paper examines the epistemic paradox whereby academia systematically excludes Wikipedia as a citable source—despite its consensus-driven editorial process, transparent revision history, and multi-tiered verification—while over-relying on traditional scholarly sources plagued by peer-review delays, reproducibility crises, and publishing monopolies. Adopting a sociology-of-knowledge framework, the study integrates bibliometric analysis and critical publishing studies to systematically analyze Wikipedia’s editorial norms, fact-checking mechanisms, and real-world citation practices, juxtaposing them against structural deficiencies in conventional academic publishing. Its primary contribution is to identify this exclusion as rooted in outdated epistemological biases; it demonstrates that Wikipedia offers context-specific cognitive advantages in timeliness, transparency, and collective verification efficiency. The paper thus argues for reconceptualizing credibility standards and developing dynamic, context-sensitive citation norms that reflect contemporary knowledge production ecologies.

0 citationsRead paper

Recent publications

Latest Papers

The tragedy of the cognitive commons: collective intelligence beyond AI-induced knowledge collapse

Jul 14, 2026

This study proposes the concept of a “tragedy of the cognitive commons” to elucidate how artificial intelligence agents may trigger a self-reinforcing degradation of humanity’s shared knowledge base—termed “knowledge collapse.” When AI substitutes for individual contributions without adequately reconstructing reliable public signals, systems risk converging to a low-knowledge equilibrium. Integrating dynamic modeling, commons theory, and empirical evidence—including a 25% decline in knowledge sharing on platforms like Stack Overflow—the paper systematically examines the complementarity between general and contextual knowledge and the associated learning externalities. The authors offer five structural critiques of prevailing models, delineate their limitations, and advance a research agenda centered on effort elasticity and governance of knowledge commons, thereby providing a theoretical foundation for improving mechanisms of human knowledge aggregation.

0 citationsRead paper

Biodiversity Media Narratives and Stock Market Performance: Evidence from Europe

Jun 18, 2026

This study investigates whether and how media coverage related to biodiversity influences stock market performance in Europe. Drawing on GDELT global knowledge graph data from France, Germany, Italy, and Spain between 2015 and 2025, the authors construct a media-based biodiversity risk index and employ a panel Granger causality test alongside an augmented inverse probability weighting (AIPW) event study design, complemented by quantile robustness checks. The analysis provides the first empirical evidence that European equity valuations are significantly driven by biodiversity-related media narratives: negative sentiment depresses stock prices, with peak effects emerging three to ten months after the initial shock. Notably, the positive impact of low-risk events substantially outweighs the negative impact of high-risk events, revealing a pronounced asymmetry. These findings remain robust after controlling for market volatility and policy uncertainty.

0 citationsRead paper

Humanity in the Age of AI: Reassessing 2025's Existential-Risk Narratives

Dec 01, 2025

This paper critically examines the empirical foundations of the “superintelligence-induced human extinction within a decade” claim promoted by two bestselling 2025 works—*AI 2027* and *If Anyone Builds It, Everyone Dies*. Method: Drawing on critical technology studies, empirical analysis of generative AI trends (2023–2025), and structural economic observations regarding compute and capital concentration—and integrating Whittaker’s algorithmic accountability and Zuboff’s surveillance capitalism frameworks—the study systematically evaluates the “intelligence explosion → superintelligence → goal misalignment” causal chain. Contribution/Results: Findings indicate that contemporary large language models remain narrow, statistical tools lacking autonomous evolution or runaway capability. The “existential risk” narrative functions less as a technical forecast than as an ideological apparatus legitimizing surveillance capitalism expansion and compute monopolization, deeply entangled with the 2025 AI financial speculation bubble. This work is the first to identify AI doomsday discourse as a capital-driven ideological construct rather than a falsifiable technological prediction.

0 citationsRead paper

Can we cite Wikipedia? What if Wikipedia was more reliable than its detractors ?

Sep 02, 2025

This paper examines the epistemic paradox whereby academia systematically excludes Wikipedia as a citable source—despite its consensus-driven editorial process, transparent revision history, and multi-tiered verification—while over-relying on traditional scholarly sources plagued by peer-review delays, reproducibility crises, and publishing monopolies. Adopting a sociology-of-knowledge framework, the study integrates bibliometric analysis and critical publishing studies to systematically analyze Wikipedia’s editorial norms, fact-checking mechanisms, and real-world citation practices, juxtaposing them against structural deficiencies in conventional academic publishing. Its primary contribution is to identify this exclusion as rooted in outdated epistemological biases; it demonstrates that Wikipedia offers context-specific cognitive advantages in timeliness, transparency, and collective verification efficiency. The paper thus argues for reconceptualizing credibility standards and developing dynamic, context-sensitive citation norms that reflect contemporary knowledge production ecologies.

0 citationsRead paper

Agentic AI Frameworks: Architectures, Protocols, and Design Challenges

Aug 13, 2025

This work addresses the lack of systematic comparison and classification frameworks for mainstream Agentic AI systems. We conduct a structured literature review and cross-framework analysis of CrewAI, LangGraph, AutoGen, and related platforms, focusing on architecture design, inter-agent communication mechanisms (e.g., CNP, A2A, ANP), memory management, security provisions, and alignment with service-oriented computing paradigms. Our method yields the first layered taxonomy for Agentic AI systems, explicitly identifying scalability, robustness, and interoperability as core bottlenecks. Innovatively, we elevate communication protocols to a primary taxonomic dimension—revealing three persistent open challenges: protocol heterogeneity, inconsistent context propagation, and weak service-contract enforcement. Based on this analysis, we propose a novel research direction: service-oriented agent collaboration protocols. This contribution provides both theoretical foundations and practical guidelines for standardizing and engineering autonomous AI systems. (149 words)

0 citationsRead paper