German Physical Society (DPG) - Working Group on Equal Opportunities (AKC)

The German Physical Society (DPG), with more than 60,000 members worldwide, is the largest physical society globally. Its Working Group on Equal Opportunities (AKC) advances gender equality, diversity, and inclusion in science and technology, providing evidence-based insights at the intersection of governance, emerging technologies, and equitable participation. Based on our hands-on experience, we propose three tangible outcomes: First, a global commitment to gender-disaggregated data collection. No international framework tracks how many women enter, stay, or leave AI-related fields. In our survey of 264 technical experts, we found that those who personally experience bias are also the most aware of how technology risks widening inequalities. Without systematic data, their experience remains invisible. The Dialogue should establish simple reporting standards for workforce diversity. What gets measured gets managed. Second, targeted funding for open-source educational resources. Our own data shows that those who feel the skills shortage most acutely point to open tools as the solution. Participants told us: open source lowers the entry barrier. The Dialogue should recommend real funding behind open education as a primary intervention for capacity-building. Third, a framework for sustainable and equitable AI infrastructure. AI depends on energy-intensive data centers, specialized hardware, and global supply chains. These realities risk concentrating AI's benefits in wealthy nations and large corporations, deepening existing divides. The Dialogue should establish principles for assessing the full lifecycle of AI systems – including energy use, resource efficiency, and equitable access to infrastructure. Sustainability must be embedded as a core governance principle, ensuring AI serves shared prosperity. A successful Dialogue delivers a roadmap with deadlines, responsibilities, and indicators. Civil society actors with implementation experience must be actively involved.

• AI capacity-building
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Open-source software, open data and open AI models

Our selected priorities reflect both UN-defined governance objectives and empirical evidence from scientific and educational practice. AI capacity-building is essential to address structural inequalities in access to technology. A/RES/79/325 explicitly highlights the need to bridge AI capacity gaps and support skills development (para. 4(b)), while the Global Digital Compact emphasizes closing digital divides and strengthening digital skills (A/79/L.2, paras. 2, 12-13). Our survey on emerging technologies confirms that lack of skilled personnel and high entry barriers are among the most significant obstacles to adoption, particularly for non-expert communities. The social, economic, ethical, cultural, linguistic, and technical implications of AI are central to ensuring that AI systems contribute positively to society. This aligns with A/RES/79/325 (para. 4(c)) and the Compact's commitment to human rights, inclusion, and equitable benefit-sharing (A/79/L.2, paras. 3, 8(c), 8(f)). Our data show a strong awareness that emerging technologies may exacerbate societal inequalities, reinforcing the need to integrate these dimensions into governance frameworks. Finally, open-source software, open data, and open AI models are key enablers of accessibility and innovation. Both A/RES/79/325 (para. 4(g)) and the Global Digital Compact (A/79/L.2, paras. 14-17) recognize their role in fostering inclusive digital ecosystems. Our findings demonstrate that open-source tools and education are perceived as the most effective mechanisms to lower entry barriers, particularly in response to skills shortages. Together, these priorities support a governance approach that is inclusive, evidence-based, and oriented toward equitable technological participation.

Yes. We identify three cross-cutting issues that are not sufficiently captured. First, "the innovation-to-adoption gap". Scientific progress is advancing rapidly, but adoption is hindered by high costs, limited practical applications, and barriers to access. Our survey found that proof-of-concept success stories are seen as the most critical accelerator. Governance must address pathways for demonstrating tangible societal value - not just supporting research. Second, "structural inequality within innovation ecosystems". This is not only an outcome of technology but is embedded in its development processes. Our data reveals a strong correlation between awareness of societal inequality and recognition of gender disparities in technical environments. Outreach data further shows that 22% of students perceive unequal opportunities, with 60% attributing this to gender-based prejudice. Those who experience bias understand the risk - but their voices are rarely included in governance discussions. Third, "the need for participatory and inclusive design". Underrepresented groups must be involved not only as beneficiaries but as active contributors to AI development and governance. This aligns with the multi-stakeholder principles of A/RES/79/325 and the Global Digital Compact (A/79/L.2). These issues are interconnected. Addressing the adoption gap requires tackling structural inequality, which in turn demands genuine participatory design. Governance frameworks must therefore move beyond principles to concrete mechanisms for inclusion at every stage - from research and development to deployment and oversight.

Governance gaps in AI capacity-building, societal implications, and open AI ecosystems are already having tangible impacts on our sector and region. A primary challenge is "unequal access to skills and infrastructure", which limits participation in AI development and application. This reflects the capacity gaps identified in A/RES/79/325 (para. 4(b)) and the persistent digital divides highlighted in the Global Digital Compact (A/79/L.2, para. 2). Our emerging technologies survey confirms that "lack of skilled personnel and high entry barriers are among the most significant obstacles", particularly for non-expert communities, reinforcing the risk of exclusion from emerging innovation ecosystems. A second challenge concerns "the societal and ethical implications of AI", particularly the risk of amplifying existing inequalities. Survey results indicate a strong perception that emerging technologies may "exacerbate societal disparities", while outreach data show that 22% of students perceive unequal opportunities, with 60% attributing this to gender-based prejudice. This demonstrates that structural inequalities—especially gender-based—persist early in education and may be reinforced by technological systems if not addressed. A third challenge is the "limited accessibility of AI tools and knowledge", linked to insufficient adoption of open-source and open-data approaches. While A/RES/79/325 (para. 4(g)) and A/79/L.2 (paras. 14–17) emphasize their importance, implementation remains uneven. At the same time, significant opportunities exist. Our data show strong support for "open-source tools and education as effective mechanisms to lower barriers", alongside the importance of "proof-of-concept applications" to demonstrate value and build trust. Strengthening these areas can foster "inclusive innovation ecosystems", enabling broader participation and more equitable distribution of AI benefits.

The Global Dialogue on AI Governance can play a critical role as a coordination and alignment platform in a fragmented and rapidly evolving landscape, where technological development is outpacing regulatory capacity and is shaped by economic competition and concentration of capabilities. AI systems are increasingly developed and controlled by a limited number of actors across a small group of countries, creating structural asymmetries in access, influence, and benefit distribution. In this context, international cooperation is necessary to prevent further fragmentation and to ensure that governance frameworks do not reinforce existing inequalities. This aligns with the Dialogue's mandate to foster inclusive, multi-stakeholder engagement and address capacity gaps (A/RES/79/325, para. 4; A/79/L.2, paras. 2, 50–52). At the same time, global governance processes risk remaining at the level of high-level principles without effective implementation, particularly where incentives favour flexibility and strategic advantage—creating the risk of "coordination without constraint." The added value of the AI Dialogue lies in its ability to: • Bridge fragmented governance approaches and promote interoperability (A/RES/79/325, para. 4(d); A/79/L.2, para. 55(b)) • Connect scientific evidence, policy, and societal perspectives, including underrepresented voices • Increase transparency around governance gaps and structural inequalities, including access, skills, and participation As a large scientific society with a dedicated structure for equal opportunities, we contribute by linking empirical evidence from research and education with societal perspectives, and amplifying underrepresented voices, particularly women and early-career scientists. To be effective, the Dialogue should support evidence-based, practice-oriented cooperation, including capacity-building and open access approaches. In this sense, it can act as a catalyst for more inclusive and coordinated global AI governance.

A wide range of initiatives already contribute to AI governance, but they remain highly fragmented across institutional, regional, and sectoral boundaries, limiting coherence and global interoperability. The AI Dialogue should therefore build on and connect these efforts to enable more integrated and inclusive governance. At the multilateral level, frameworks such as the Global Digital Compact and the mechanisms established under United Nations Resolution 79/325, including the Independent International Scientific Panel on AI, provide a foundation for evidence-based and inclusive governance. These can be complemented by multi-stakeholder platforms such as the Internet Governance Forum and standard-setting and policy work led by organizations such as the OECD and UNESCO. At regional and national levels, initiatives such as the European Union's AI regulatory framework and national AI strategies coexist with scientific and professional networks, which contribute domain expertise and education-oriented engagement. However, the current landscape demonstrates that AI governance cannot be effectively advanced through parallel or purely top-down approaches. Instead, it requires structured interaction between policy, science, industry, and civil society. The absence of such coordination risks duplication of efforts, uneven access to knowledge, and persistent exclusion of underrepresented groups. The added value of the AI Dialogue lies in its role as a "system-level bridge function": enhancing interoperability between existing initiatives, connecting scientific evidence with policy development, and translating governance principles into practice-oriented cooperation. It can further support capacity-building and the development of more open and accessible AI ecosystems. In this context, the Working Group on Equal Opportunities of the German Physical Society (AKC) provides a relevant complement by linking scientific expertise with education and outreach, and by contributing evidence on structural barriers—particularly in skills development, access, and gender equality—that are essential for inclusive AI governance.

Different stakeholders can contribute to the AI Dialogue through clearly defined but interconnected roles reflecting the multi-dimensional nature of AI governance, spanning science, policy, industry, and civil society. To ensure meaningful impact, the Dialogue should be structured as a co-creation platform rather than a purely consultative forum, enabling iterative development of governance approaches grounded in evidence and lived experience. Scientific and technical communities can provide evidence-based analysis of AI capabilities, risks, and limitations, while industry contributes implementation perspectives. Governments ensure regulatory coherence, and civil society brings societal accountability and legitimacy. However, experience from previous digital and technology governance processes indicates that when participation is dominated by institutional and high-level actors alone, critical perspectives related to access, education, and structural inequality—particularly gender inequality—remain insufficiently integrated, limiting both inclusiveness and long-term effectiveness. The Working Group on Equal Opportunities of the German Physical Society (AKC) exemplifies the added value of intermediary scientific-societal actors. Such actors connect research, education, and outreach, and provide empirical insight into structural barriers in AI ecosystems, including gender disparities in STEM pathways, unequal access to skills development, and early-stage attrition of women in technical fields. Without such perspectives, governance frameworks risk overlooking foundational conditions required for equitable participation in AI development and deployment. To strengthen the Dialogue, it should: • establish thematic working tracks integrating technical, ethical, and socio-economic dimensions • ensure systematic inclusion of gender and equality-focused scientific actors, not only institutional representatives • embed education and workforce development perspectives as core inputs, not peripheral consultations • create iterative feedback mechanisms linking policy outputs with empirical evidence from practice In this structure, the AI Dialogue can move beyond dialogue alone and function as a co-creation mechanism for inclusive, gender-responsive, and evidence-based global AI governance.

Global discussions on AI governance continue to be dominated by actors from a limited number of regions, institutions, and large-scale industry and policy bodies. As a result, several critical perspectives remain structurally underrepresented. First, voices from low- and middle-income countries are often insufficiently integrated, despite being disproportionately affected by unequal access to digital infrastructure and AI capabilities. Second, education and workforce development actors, including teachers, curriculum designers, and STEM outreach practitioners, are rarely systematically included, even though they determine access pathways into AI-related fields. Third, early-career researchers and practitioners are underrepresented compared to established institutional experts, despite being directly involved in implementation realities. Finally, gender-focused and equity-oriented scientific perspectives remain fragmented rather than structurally embedded in governance processes. These gaps are not only representational but also affect the effectiveness of governance, as they limit understanding of how AI systems interact with education systems, labor markets, and social structures. The Working Group on Equal Opportunities of the German Physical Society (AKC) illustrates how these gaps can be addressed in practice. As a societal body combining research expertise with education and outreach activities, AKC brings empirical insight into gender disparities in STEM pathways, access barriers in technical education, and structural inequalities in scientific careers. Such intermediary actors can translate between policy, research, and lived educational realities. Inclusion can be strengthened through structured mechanisms such as: • structured participation of education- and equity-focused actors • regional and early-career advisory channels • integration of gender-disaggregated and access-related evidence • engagement formats beyond capital- and institution-centric representation In this way, the AI Dialogue can become more inclusive, evidence-informed, and representative of the full AI ecosystem.

To move beyond traditional consultation formats, the AI Dialogue should be designed as a working interaction space where stakeholders jointly construct understanding and solutions, rather than merely exchange positions. A first useful format would be problem-focused "co-creation tracks", where diverse stakeholder groups—policy, science, industry, and civil society—work on one concrete governance challenge over multiple sessions. This would allow continuity, iteration, and a shift from abstract principles to practical trade-offs. Second, the Dialogue could introduce structured "perspective rotation sessions", where participants are explicitly asked to switch roles—for example, policymakers temporarily evaluate from the perspective of educators, or industry representatives from the perspective of early-career researchers. This helps reveal blind spots that typically remain hidden in static formats. Third, evidence-to-policy translation labs could be established, where empirical findings (e.g. from surveys, education systems, or sectoral studies) are directly mapped into governance implications in real time, ensuring that data is not only presented but actively used in shaping recommendations. The Working Group on Equal Opportunities of the German Physical Society (AKC) could contribute particularly through a "science–education interface format", bringing together researchers, educators, and inclusion experts to explicitly examine how AI governance decisions affect access to skills, gender equality in STEM pathways, and long-term participation in AI ecosystems. This is crucial because such dimensions are often discussed separately from technical governance, despite being structurally linked. Finally, the Dialogue should include short iterative feedback cycles rather than one-off consultations, ensuring that outcomes are continuously refined and validated by different stakeholder groups. Overall, these formats would transform the AI Dialogue into a dynamic co-production space, strengthening both legitimacy and practical relevance of global AI governance.

Effective AI governance increasingly emerges from a combination of regulatory frameworks, open knowledge infrastructures, and education- and evidence-based implementation approaches rather than single-point technical solutions. At the policy level, the Global Digital Compact and United Nations Resolution 79/325 provide important examples of multi-stakeholder and capacity-oriented governance approaches. Similarly, the European Union's AI regulatory framework demonstrates how binding rules can be combined with risk-based approaches and implementation guidance to ensure both innovation and accountability. At the platform level, the Internet Governance Forum represents a long-standing model for inclusive, multi-stakeholder dialogue that connects technical communities, governments, civil society, and academia without a hierarchical structure. From a practice perspective, open-source and open-data ecosystems are increasingly recognized as key enablers of transparent and accountable AI development, particularly when combined with educational initiatives that lower entry barriers and support capacity-building across regions. In this context, scientific societies such as the German Physical Society provide an important complementary governance function through education, training, and evidence generation. Within this structure, the Working Group on Equal Opportunities (AKC) contributes by promoting inclusive STEM participation, identifying structural barriers in access to scientific education, and supporting gender equality in technical fields. These activities are indirectly but concretely relevant to AI governance, as they address the foundational conditions required for broad participation in AI development and oversight. Overall, effective AI governance is best supported by layered approaches combining regulation, open knowledge systems, and inclusive education infrastructures, ensuring that governance is not only norm-setting but also implementation-capable and socially grounded.