AURORA - International Study Center, a research center focusing on innovation, neurosciences, cognitive diversity and peace

A key outcome would be the recognition and adoption of a Cognitive Interoperability Registry combined with a Dynamic Certification system for artificial intelligence systems. These tools are designed to ensure that AI can operate effectively in real-world contexts, adapting to diverse cognitive profiles and cultural backgrounds. The registry collects structured information on AI systems, including interaction models, adaptation capabilities to specific cognitive patterns, learning and data-update mechanisms, and the cultural and operational contexts in which the AI is tested. The technical precision of the registry allows for standardization of key attributes, such as the ability to recognize and adapt to different cognitive patterns, management of cognitive biases, resilience to interpretive errors, and compatibility with intercultural interaction protocols. The Dynamic Certification complements the registry by providing a system of continuous evaluation. Each registered AI is tested in controlled simulations that replicate cognitively complex scenarios, involving real users representing diverse cognitive profiles and cultural contexts. The certification is periodically updated based on observed results, measuring quantitative metrics such as the percentage of correct responses in standardized scenarios, the average adaptation time to new cognitive profiles, the capacity to manage cognitive conflicts, and the degree of generalization of responses across different contexts. This approach provides policymakers with concrete tools, bridging the gap between ethics and real-world applications, and ensuring AI operates reliably and adaptively. To complete the framework, global collaborative simulations provide a testing environment where registered and certified AI are evaluated in realistic contexts, involving users from geographical areas with varied cognitive levels and cultural backgrounds In the initial phase, a pilot project allows the registry and certification system to be tested with a limited number of AI systems and stakeholders. In the next phase, the platform expands to a continental scale, integrating more complex scenarios and a larger number of participants.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Interoperability of governance approaches
• Transparency, accountability, and human oversight

The selection reflects the areas where our proposed framework, the Cognitive Interoperability Registry combined with a Dynamic Certification system, can have the most immediate and measurable impact. Safe, secure, and trustworthy AI is addressed through the Dynamic Certification system, which continuously evaluates AI performance in diverse cognitive and cultural scenarios. This ensures systems operate reliably, manage biases, and adapt to complex real-world contexts, providing policymakers with concrete evidence of safe and trustworthy AI behavior. Interoperability of governance approaches is a central focus of the Cognitive Interoperability Registry, which standardizes structured information on AI systems' cognitive adaptation, learning mechanisms, and operational performance. By providing a common reference framework, the registry enables consistent assessment across different systems and jurisdictions, facilitating coordinated and interoperable governance. Social, economic, ethical, cultural, linguistic, and technical implications of AI are addressed through global collaborative simulations, which test AI systems with diverse users representing varied cognitive profiles and cultural backgrounds. These simulations generate actionable data that informs improvements in inclusiveness, adaptability, and technical reliability, ensuring that AI deployment accounts for social and cultural diversity. Finally, Transparency, accountability, and human oversight are embedded throughout the framework. The registry and certification system provide verifiable, measurable metrics on AI performance, allowing continuous monitoring and auditability. Policymakers can track system behavior, assess compliance with standards, and identify areas for improvement, bridging the gap between high-level ethical principles and operational reality. By focusing on these four areas, our proposal offers practical, scalable, and evidence-based tools that enable policymakers, technical communities, and stakeholders to implement AI governance that is not only theoretically sound but operationally effective across diverse global contexts.

Yes. While the listed themes cover key aspects of AI governance, our experience suggests that cognitive interoperability and continuous operational certification represent emerging cross-cutting issues not yet fully captured. Current discussions focus on high-level principles, ethics, trust, safety, and human rights, but often lack tools for real-world assessment of AI behavior across diverse cognitive and cultural contexts. AI systems increasingly interact with users whose cognitive styles, cultural backgrounds, and linguistic profiles vary widely. Ensuring these systems perform reliably and inclusively requires mechanisms that measure and certify adaptive performance, rather than only relying on static standards or abstract guidelines. The concept of Cognitive Interoperability addresses this gap by standardizing how AI systems' cognitive adaptation, learning mechanisms, and operational responses are evaluated and compared. Complementing this, a Dynamic Certification system provides continuous monitoring and validation of AI behavior in realistic, heterogeneous scenarios. Together, these tools allow for measurable, evidence-based assessment of AI systems' adaptability, reliability, and inclusiveness, factors critical to trustworthy deployment yet insufficiently addressed in current frameworks. Additionally, these approaches facilitate global collaboration by enabling shared benchmarks and interoperable evaluation across jurisdictions and technical communities. This allows stakeholders, including policymakers, technical experts, and civil society, to make informed, data-driven decisions and to iteratively improve governance frameworks as AI systems evolve. In summary, cognitive interoperability and dynamic certification are emerging priorities that cut across ethics, technical performance, inclusiveness, and policy oversight. Incorporating these mechanisms into international AI governance discussions ensures that frameworks are not only principled but also operationally effective, evidence-based, and globally scalable.

In Italy, the deployment of artificial intelligence technologies highlights particularly significant challenges. Fragmentation between national, regional, and local levels makes it difficult to implement uniform governance standards, while digital divides across regions and between public and private sectors limit the consistent and reliable evaluation and adoption of AI systems. Furthermore, data sharing and interoperability between public institutions, research centers, and industry remain limited, reducing the effectiveness of governance policies and initiatives. However, these challenges also present important opportunities. Through the Cognitive Interoperability Registry and the Dynamic Certification system, developed by the Aurora International Study Center, it is possible to generate concrete operational data and verifiable metrics that assess AI reliability, adaptability, and inclusiveness in real-world contexts. These tools allow ethical principles and policy guidelines to be translated into implementable practices, bridging the gap between abstract recommendations and operational reality. Additionally, the active involvement of AURORA International Study Center and other local initiatives enables the creation of a model that is replicable at the European and global levels, fostering collaboration among public administrations, universities, independent research centers, and industry. In this way, Italy can contribute in an original and tangible way to the development of global operational standards for AI safety, reliability, and interoperability, demonstrating how a practical, evidence-based approach can support inclusive, measurable, and scalable governance at the international level.

The AI Dialogue can play a strategic role if it becomes a space where academic research, together with governments, the private sector, and civil society, contributes tangibly to the development of concrete tools for AI governance. In particular, the academic community has significant potential: the ability to produce evidence, verifiable metrics, and realistic scenarios that connect ethical principles to operational practices. The Dialogue can encourage the active participation of universities, research centers, and independent laboratories, enabling the sharing of replicable results and tools applicable in diverse contexts. In this way, the AI Dialogue becomes not only a forum for discussion but also an international laboratory for testing operational practices and standards, strengthening multilateral cooperation with verifiable and shared instruments. It can thus enhance the contribution of the academic sector by linking research, innovation, and real-world applications, providing evidence that helps governments and stakeholders make more effective decisions, promote inclusion, and develop scalable standards globally. The true value of this space lies in its ability to transform abstract principles into concrete, shared, and measurable actions, creating a tangible and lasting impact on international AI governance.

There are numerous international initiatives addressing artificial intelligence governance. The AI Dialogue can build on these foundations by providing a structured and ongoing forum in which empirical evidence, verifiable metrics and operational tools are shared among governments, civil society, the private sector and the scientific community. Aurora International Study Center contributes concretely to this dialogue through independent research published in peer‑reviewed academic journals, preparatory policy studies, and responses to public consultations initiated by the European Commission. A team of very young researchers aged 21–22 participates actively in this work alongside more senior scholars, producing analyses and publications that connect academic research with practical applications. In doing so, Aurora offers the Dialogue concrete contributions based on data and verifiable evidence, enriching international cooperation with fresh perspectives and replicable tools.

Different stakeholders can contribute more effectively to the AI Dialogue by overcoming a limitation that currently characterizes many discussions on artificial intelligence governance: the tendency to treat "AI" as a single, uniform system, without distinguishing between models, functions, and deeply divergent application contexts. While initial attempts exist to differentiate AI systems, these distinctions have not yet been systematically integrated into the structure of international governance processes. Governments, for example, could contribute by developing more granular regulatory approaches, differentiating between AI types (generative, predictive, decision-making, autonomous) and high-impact application sectors, thereby avoiding generalized approaches that risk being ineffective or excessively restrictive. Similarly, the private sector could contribute not by indiscriminately expanding use cases, but by focusing on areas where AI delivers measurable and verifiable value, avoiding patterns observed in previous emerging technologies, where widespread adoption preceded a clear understanding of appropriate applications. The academic community and research centers can contribute by clarifying these distinctions, producing comparative analyses across different AI systems, operational contexts, and outcomes, and providing interpretative tools that enable more precise governance. Experiences developed within the Aurora International Study Center, which combine independent research, scientific production, and contributions to public consultations, demonstrate how theoretical analysis can be concretely linked to operational implications. Regarding format, the AI Dialogue could be structured not only by stakeholder type but also by AI type and application context, creating focused working spaces (for example: generative AI in information systems, predictive AI in healthcare, decision-making AI in public systems). This would allow the development of more precise recommendations, more relevant metrics, and standards that are genuinely applicable. A Dialogue structured in this way can contribute to more effective international cooperation because it is based not on generalizations, but on operational differentiations, comparable evidence, and informed decisions about where and how AI should be applied.

Global discussions on AI governance increasingly involve a wide range of stakeholders, yet some important perspectives remain underrepresented—not necessarily in terms of identity, but in the nature of the contributions they bring. In particular, contributions from interdisciplinary and independent research contexts are still insufficiently valued. These actors, often operating outside large academic or industrial institutions, are well positioned to connect technical, cognitive, ethical, and operational dimensions of AI systems. They can offer flexible, experimental, and context-aware analyses, but often face barriers to meaningful participation in international decision-making processes. At the same time, perspectives that treat cognitive and cultural diversity as central variables in the design and evaluation of AI systems remain marginal. Discussions tend to focus on general principles, without sufficiently addressing how different cognitive models, linguistic structures, and interpretative frameworks concretely shape human–AI interaction. In addition, younger generations of researchers and professionals are still only partially involved, despite their active role in knowledge production and critical engagement with emerging technologies. To better include these perspectives, the AI Dialogue could adopt more open and structured participation mechanisms that value evidence-based contributions, comparative studies, and interdisciplinary approaches. This could include the direct involvement of independent research centers, dedicated spaces for addressing cognitive and cultural diversity, and the active integration of early-career researchers into consultation and co-creation processes. Expanding participation in this direction would enrich global discussions with less standardized but highly relevant perspectives, contributing to a more inclusive, informed, and context-aware approach to AI governance.

Innovative engagement formats for the AI Dialogue should move beyond traditional panel discussions and actively rethink how stakeholders interact. One particularly effective approach would be the introduction of role-exchange sessions, in which participants temporarily assume the perspectives and responsibilities of other actors. For example, policymakers could engage as private sector representatives, researchers as regulators, or industry actors from the perspective of civil society. This type of exercise helps participants better understand constraints, incentives, and decision-making processes across sectors, leading to more balanced and realistic outcomes. A second key element concerns the need to reduce implicit hierarchies within the Dialogue. Rather than reinforcing institutional or geopolitical differences, formats should be designed to place participants on equal footing, enabling more open and authentic exchange. This can be achieved through small-group working sessions, anonymized contributions, or deliberative methodologies that prioritize ideas over formal positions. In addition, the AI Dialogue could benefit from continuous engagement mechanisms, moving beyond the logic of episodic meetings. Regular working sessions, rotating participation models, and collaborative platforms would allow for broader and more diverse contributions over time, including from actors who are often excluded from more formal settings. Finally, the creation of heterogeneous discussion clusters, where stakeholders from different sectors, geographic regions, and areas of expertise are intentionally mixed, can foster more dynamic interactions and reduce the risk of siloed or self-referential approaches. Taken together, these formats can transform the AI Dialogue into a truly interactive and adaptive space, capable not only of facilitating discussion, but also of generating shared understanding and actionable outcomes.

Several approaches can contribute to more effective AI governance by strengthening the link between scientific knowledge and policy implementation. One relevant approach is the development of interdisciplinary policy labs embedded within public institutions, where multidisciplinary teams support decision-making processes in real time. By combining expertise from fields such as data science, social sciences, and cognitive research, these labs enable the integration of scientific evidence into governance, supported by operational tools such as monitoring dashboards and structured reporting mechanisms. A second promising approach involves the creation of dynamic expert networks at the national and international levels. These networks allow policymakers to access continuously updated knowledge through shared platforms, standardized analytical frameworks, and coordinated responses to emerging challenges. Such mechanisms improve the speed and quality of decision-making while fostering collaboration across institutions and geographic regions. In addition, evidence-to-policy accelerators can help translate research findings into concrete interventions. Through short-cycle experimentation, pilot testing, and iterative feedback, these mechanisms ensure that policy solutions are tested in real-world contexts before being scaled, reducing the gap between theoretical research and practical implementation. Across these approaches, a key element is the integration of ethical oversight and accountability mechanisms, including transparency tools, impact evaluation systems, and independent review processes. Taken together, these models provide concrete ways to move from principle-based discussions to operational governance, ensuring that AI-related policies are evidence-based and adaptable to complex real-world environments.