Conference Agenda

1. Background

The inherent "black box" nature of machine learning AI poses a fundamental challenge that hinders the realization of its potential. This opacity creates significant information asymmetry, leading to market distrust and potential efficiency losses that impede the widespread adoption of beneficial AI technologies. Establishing trust is widely recognized as a prerequisite for the adoption of AI systems (Schmidt et al. 2020). In this context, transparency is frequently proposed as a mechanism for fostering such trust. While the Hiroshima AI Process (HAIP) provides a borderless, voluntary reporting framework (G7 2023), its long-term effectiveness depends on whether the market recognizes and rewards such transparency through economic incentives.

2. Research Question

This study empirically examines whether specific AI governance "articulations" under the HAIP translate into tangible market rewards. If information disclosure through the HAIP has the power to increase market-based rewards, corporate participation can be expected to be long-term sustainable, as it would be aligned with profit-maximization motives. To verify this, we address the following two-step inquiry: (1) To what extent do organizations differentiate their governance narratives within the flexible, open-ended HAIP template, and what distinct "articulation styles" emerge from these disclosures? (2) Does the "Japanese Style" of governance—identified as a prominent, rule-based, and institutionally embedded approach in our Stage 1 analysis—lead to higher consumer trust and Willingness-to-Pay (WTP) compared to other styles?

3. Methodology

• Stage 1: Computational Text Analysis (Completed). We analyzed 26 AI governance reports under the HAIP framework. Using Term Frequency-Inverse Document Frequency (TF-IDF) vectorization and Uniform Manifold Approximation and Projection (UMAP), we successfully extracted clear structural differentiations in governance narratives. Crucially, this analysis identified a distinct "Japanese Style" cluster, capturing a specific, principle-driven, and rule-based organizational philosophy prevalent among top Japanese firms, distinct from the model-centric narratives of global providers. These findings are well compatible with the outcomes of authors' interviews with HAIP-participating firms (Author et al. 2025), which provided foundational evidence of the diverse governance philosophies and implementation challenges among global AI actors.
• Stage 2: Consumer Choice Experiment (In Progress). Built upon the evidence from Stage 1, we designed a consumer survey (n = approx. 1,000, scheduled for April 2026) using the Contingent Valuation Method (CVM). The experiment utilizes these real-world articulation styles to compare "Technical Control-centric" vs. "Japanese-style Organizational/Rule-centric" governance in medical AI scenarios to measure the impact on consumers' WTP and trust recovery.

4. Novelty and Expected Results

This study moves beyond theoretical discussions of AI transparency by examining whether voluntary governance disclosures can generate tangible market incentives. While prior studies indicate that trust is a key prerequisite for the adoption of AI systems and that transparency-related attributes influence user evaluations and reliance on AI (Schmidt et al. 2020), empirical evidence on whether governance disclosures themselves function as economically meaningful market signals remains limited. Our Stage 1 analysis has already identified a distinct “Japanese Style” of AI governance characterized by rule-based articulation and institutionally embedded frameworks, contrasting with the model-centric narratives often emphasized by global AI providers. Stage 2 tests whether such institutionalized governance articulation can influence consumer trust and WTP in AI services. Building on previous studies showing that consumers respond to transparency and explainability attributes when evaluating AI systems (Ioku et al. 2024; König et al. 2022), and that disclosure of AI involvement shapes perceptions of authenticity and trust-related evaluations (Kučinskas 2025), the experiment examines whether governance disclosures framed in institutional and rule-based terms can function as credible signals in consumer decision-making. If such articulation generates measurable WTP premiums, it would suggest that transparency frameworks such as those promoted under the HAIP may operate not only as governance norms but also as economically meaningful mechanisms sustaining voluntary compliance regimes.

5. Policy Implications

This study provides a strategic empirical basis for the design of global AI governance, addressing the transition from voluntary "soft law" to mandatory "hard law." If the identified "Japanese Style" of governance commands a WTP premium, it validates the sustainability of voluntary regimes like the HAIP as self-regulating market infrastructures, suggesting that policymakers should focus on standardizing disclosure templates to further lower transaction costs. Conversely, if the market fails to reward such rigor, it signals a fundamental market failure in valuing non-technical safeguards, necessitating more interventionist shifts such as integrating AI governance audits into public procurement standards or establishing mandatory disclosure requirements. Ultimately, this study's findings offer an evidence-based roadmap for balancing compliance burdens with market-driven innovation.