CeADAR

Clear roadmap and recommendations for addressing trade-offs when implementing AI policies

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

Currently, there's a polarisation in society regarding the use of AI: enthusiastic vs. cautionary. To leverage the potential of these technologies in a sustainable way, a balanced approach is needed. Promoting responsible use and data literacy, not only capacitation (AI capacity-building can be understood as how to promote use, but it's more important when and why). A long-term strategy must consider a broad vision of implications, not only direct providers, deployers, and users, but also indirectly impacted people. Analysis of the impact on quality of life, and not only outcomes generation, must be considered.

- AI Literacy vs. Capacitation: AI capacity-building is confusing. It is important the difference between AI literacy (critical use, understanding capabilities, risks and threats) needed for democratic resilience and AI capacitation (learning how to implement or use AI). Mixing them up leads to "one-size-fits-all" policies that fail both goals. - Geopolitics & Dependencies: Many nations fear a new form of colonialism where they provide the data/labor while a few corporations/nations own the power and "intelligence." - Environmental Costs: water and energy footprints of LLMs are massive hurdles for the Sustainable Development Goals (SDGs) - Ensuring AI-Security/Access as some AI systems are becoming commodities. If AI becomes as essential as electricity, then "denial of service" or restricted access becomes a human rights and economic security issue

Current market dynamics, driven by multinational urgency, have fostered a climate of "innovation anxiety." This has led to a hidden extractive model where workers, distracted by the pressure to integrate AI, inadvertently outsource their unique expertise to train centralized proprietary systems. Without governance to protect knowledge sovereignty, this trend risks devaluing human labor and creating a cycle of dependency on a few global entities. While AI offers significant efficiency gains, current developments are disproportionately focused on "dopamine-rewarding" interfaces designed for immediate user engagement. To ensure sustainable integration, governance must pivot from prioritizing short-term outcome generation to fostering long-term strategies that measure success through the enhancement of quality of life and the protection of indirectly impacted populations.

Advocate for a "Right to Knowledge Attribution" to prevent extractive training models. Propose "Quality of Life" as a standard KPI for AI deployment instead of only "Outcome Generation" of most benchmarks. Promote international guidelines against "dopamine-rewarding" systems in professional AI tools and social media platforms to reduce pernicious side effects and worker distraction.

Initiatives that are in contact with small businesses and the civil sector. In Europe, the European Digital Innovation Hub network provides access to regional hubs supporting companies, especially SMEs. GlobAIpol is a global network for AI Policy Research

Adopt a decentralized "Hub-and-Spoke" structure that leverages networks like Europe's EDIHs to funnel real-world SME concerns and "friction reports" directly into global policy. With over 200 hubs, EDIHs act as "One-Stop Shops" that translate local business reality into policy recommendations. At least 50% of floor time reserved for non-governmental and non-multinational companies' voices. Incorporate random-sample citizen juries. Adopt a structure that prioritises hyper-local inputs, through civil society and trade unions

The Global Majority or Global South. Individuals indirectly impacted by AI, like traditional artisans and artists whose livelihoods are displaced by generative AI, or citizens in countries with weak labor laws where AI is used for surveillance or automated firing. SMEs and local innovators, whose perspective on vendor lock-in and the need for open-source public infrastructure is often drowned out by Big Tech lobbying, sometimes even unconsciously. At least 50% of floor time reserved for non-governmental and non-multinational companies' voices. Incorporate random-sample citizen juries. Adopt a structure that prioritises hyper-local inputs, through individuals, civil society and trade unions.

Prioritise the voice of the underrepresented and let them speak first, as this shifts the point of view. Global campaigns using a multichannel hyperlocal approach. Use multiplier structures, such as the EDIH network, commerce chambers, government and municipalities, global associations...

EDIH Network https://european-digital-innovation-hubs.ec.europa.eu/home Globaipol https://www.linkedin.com/company/globaipol