Independent Chronicler | The Royal Golden Cocoon of Java: Cricula trifenestrata (Indonesia)

A successful first Global Dialogue on AI Governance would be one that moves from principles to evidence-based, rights-based governance grounded in real-world cases, recognizing that AI systems can misclassify and erase significant knowledge when data is incomplete or biased. It should prioritize closing digital and data divides, ensuring equitable representation of underrepresented knowledge systems, and establishing mechanisms for continuous data updating and scientific accuracy, as outdated or incorrect data can create systemic errors that lead to development stagnation, systemic jeopardy, and the erosion of the right to development. The Dialogue should embed inclusive, human rights–based frameworks that integrate local expertise, cultural context, and open science principles, supported by transparency and human oversight. Ultimately, success would mean ensuring that AI systems do not reinforce inequities, but instead uphold cultural rights, enable fair participation, and support inclusive, accurate, and accountable development pathways.

• Protection and promotion of human rights
• Transparency, accountability, and human oversight
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Safe, secure and trustworthy AI

The selected priorities reflect the need to address how AI systems can amplify biased or incomplete data from input to output, leading to misclassification and the erasure of culturally and scientifically significant knowledge. Ensuring safe, secure and trustworthy AI requires improving data quality and preventing the reinforcement of existing inaccuracies. Protection and promotion of human rights is essential, as such biases undermine cultural rights and the right to development. Transparency, accountability, and human oversight are needed to identify, correct, and continuously update data to avoid systemic errors. Finally, the social, economic, cultural, and linguistic implications of AI must be addressed, as biased outputs can lead to exclusion, misrepresentation, and inequitable development outcomes.

A key cross-cutting issue not fully captured is how data is indexed, prioritized, and interpreted within AI systems, which directly shapes outputs. Much of the data used by current AI models is drawn from online environments where visibility, frequency, and repetition influence what is recognized as valid input. Rare or underrepresented terms are often misinterpreted or replaced by more common ones through pattern recognition processes-for example, "Cricula," a wild silkworm with cultural and ecological value that has been widely misclassified as a "pest," being read as "circular"-leading to distortion at the input stage. These biases are then amplified from input to output, reinforcing misclassification and disinformation at scale. This creates systemic risks, particularly for culturally and scientifically significant knowledge that is underrepresented online, resulting in invisibility, misrecognition, and barriers to development. Addressing this requires governance approaches that ensure not only transparency, but also data accuracy, contextual understanding, and continuous updating, so that knowledge systems reflect what is correct and current, rather than what is most visible or frequent.

Governance gaps in AI are affecting the region through misclassification and data invisibility due to lack of accurate online data, where systems rely on incomplete or biased inputs. In Indonesia, Cricula trifenestrata, a wild silkworm with ecological, health, and cultural value, has been widely misclassified as a "pest." Such cases show how AI can amplify errors from input to output, shaping perception, limiting recognition, and hindering scaling and global development opportunities. The key challenge is that underrepresented knowledge systems remain invisible or mischaracterized, leading to inequitable outcomes. The opportunity lies in strengthening transparency, human oversight, and continuous data updating, ensuring that AI systems reflect accurate and current knowledge and support more inclusive, scalable, and fair development pathways.

The AI Dialogue can advance international cooperation by linking AI governance to existing global frameworks, rather than operating in isolation. It should align with instruments such as the Convention on Biological Diversity and the Kunming-Montreal Global Biodiversity Framework (KM GBF), as well as initiatives under the United Nations Decade on Ecosystem Restoration, ensuring that AI supports ongoing efforts to address urgent challenges like biodiversity loss and climate change. By grounding discussions in real-world cases, the Dialogue can help harmonize approaches on data accuracy, continuous updating, and human oversight, while enabling more inclusive participation from underrepresented regions. In this way, it can strengthen coordination across sectors and ensure AI governance contributes to, rather than fragments, global sustainability and development agendas.

The AI Dialogue should build on existing initiatives such as the Global Digital Compact and frameworks under the United Nations system, including the Convention on Biological Diversity and the Kunming-Montreal Global Biodiversity Framework. These already provide direction for addressing urgent challenges like biodiversity loss and climate change. The added value of the Dialogue is to bridge fragmentation, aligning efforts across sectors and countries, and ensuring that AI governance supports the implementation and monitoring of these frameworks rather than operating in isolation.

Inclusive participation can be strengthened by continuing the current approach, where different stakeholders contribute based on their roles—governments on policy, civil society and independent experts on real-world cases, and technical communities on system design. The AI Dialogue is already moving in this direction, and this should be further supported through case-based discussions, hybrid participation, multilingual inputs, and open submission channels. Strengthening mechanisms for synthesis and follow-up will ensure that contributions are translated into actionable outcomes, maintaining an inclusive, grounded, and collaborative process.

Underrepresented voices in global AI governance include Global South communities, local practitioners, and non-dominant language speakers, whose knowledge is often not accurately captured or translated in digital systems. For example, Javanese knowledge can become unreadable or incorrectly interpreted, creating barriers to visibility and participation. This language gap limits how local initiatives are understood within global frameworks, hindering fair access and equitable benefit sharing, even when the knowledge itself is valuable. Inclusion can be strengthened by improving multilingual support, accurate translation, and contextual interpretation, alongside open participation channels. Creating bridges between local knowledge systems and the language of global frameworks is essential, so that underrepresented communities can contribute effectively and have their knowledge recognized, integrated, and fairly valued within AI governance and development processes.

Innovative engagement can be strengthened by broadening the range of participation within existing stakeholder groups, particularly by including underrepresented private and commercial actors, beyond the usual policy and technical focus. As emphasized in the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services Business and Biodiversity Assessment, real-world uptake depends on engaging the private sector. This should extend to knowledge translation channels such as publishers and creative industries, which play a key role in bringing knowledge into visibility, markets, and public understanding.

Effective AI governance can build on open, participatory models such as the external review process of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services, where experts and stakeholders globally contribute to assessments through transparent and inclusive review. This approach ensures that knowledge is open, scrutinized, continuously improved, and not erased, helping to build trust in both data and outcomes, which is essential for reliable AI systems. Such practices can be complemented by open science principles, where data is published, accessible, and regularly updated, reducing the risk of outdated or biased information being amplified by AI systems. Combined with governance approaches that emphasize transparency, accountability, and human oversight, these models offer practical ways to ensure AI systems remain accurate, inclusive, and trustworthy over time.