FST JOURNAL

The session was led by Zoi Roupakia, a Policy Affiliate at Cambridge Industrial Innovation Policy (Cambridge University), CAIDP Policy Group Lead, and Founder of Noetic AI. She examined how AI ecosystems can support sustainable and inclusive growth. Drawing on global survey data and governance indices, Zoi explored links between public trust, governance, and AI readiness. Evidence showed that optimism toward AI is higher in countries with strong governance frameworks, responsible AI capabilities, and robust data infrastructure. Structural challenges included digital divides, compute concentration, and uneven global distribution of AI capabilities.

 Within this context, the UK was characterised by fragile public attitudes toward AI, uneven business adoption concentrated among larger firms and specialised sectors, and a relatively small share of global high-performance computing capacity despite strong research strengths.

Following the presentation, participants discussed implications for science, society, and innovation, focusing on trust, governance, and policy implementation. Although awareness of AI is increasing globally, concern continues to outweigh excitement. Trust varies by application, and effective governance was seen as central to public confidence.

 Key themes

 Discussion of the UK focused on public attitudes, business adoption, and infrastructure. It was mentioned that growing familiarity with AI has led to greater scrutiny, while adoption remains uneven. Participants highlighted challenges related to access to compute resources, open infrastructure, and long-term competitiveness.

 It was also emphasised that responsible AI depends on governance in practice, not policy design alone. Gaps between policy ambition and implementation risk slowing innovation without preventing harm, while uneven access to skills, infrastructure, and digital literacy may reinforce existing inequalities.

 Skills, education, and productivity were key themes. AI has strong potential to enhance productivity, but benefits are not automatic. Upskilling citizens, strengthening digital literacy, and promoting adaptable learning were viewed as essential, particularly as AI discussions in universities have yet to reach wider public audiences.

 Ethical and environmental concerns included surveillance risks from computer vision, large-scale data collection, public misunderstanding caused by broad use of the term “AI,” and the environmental impact of large language models alongside limited benchmarking practices.

 Looking ahead, participants pointed at priorities including specialised domain-specific systems, continuous learning models, and improved explainability to strengthen transparency and trust. Policy directions aligned with the UNECE Transformative Innovation Policy Charter included mission-oriented investment, experimentation-based governance, strategic public procurement, international collaboration on standards, and institutional capacity-building.

 Overall, the session emphasised that responsible AI requires sustained attention to governance implementation, skills and education, transparency, infrastructure access, and public engagement to ensure societal benefit.

 Recommendations for stakeholders

 Strengthen governance implementation alongside policy design. Translate AI policy ambitions into effective operational governance through experimentation-based approaches such as regulatory sandboxes and learning-based oversight.

Invest in skills, digital literacy, and institutional capacity. Expand workforce development and education focused on adaptability and lifelong learning to ensure AI productivity gains are broadly shared.

 Improve access to infrastructure and promote trustworthy AI ecosystems. Address compute concentration and uneven infrastructure access while supporting strategic public procurement and international collaboration on shared standards.