Decoding the Global Governance Landscape
In the constantly evolving world of artificial intelligence (AI), global governance has become a multi-national endeavor. As the OECD Policy Observatory informs, there are currently 668 national governance initiatives from 69 countries across the globe. This extensive list comprises national strategies, agendas, projected plans and joint initiatives aimed at shaping AI in the public sector. On the other hand, the OECD also notes 337 enforcement initiatives pertaining to AI regulations and standards.
Governance, in its AI setting, encompasses a broad array of insights. It includes the safety parameters of AI tools, policies on data access, model usage, and the implemented regulations. This is why the AI governance process must start at the conceptual level and run throughout the AI solution’s lifecycle.
Common Challenges in the Pursuit of Governance
In practice, government agencies strive for governance that caters to economic prosperity, national security, and political dynamics. A prime example is the recent directive by the U.S. White House order to establish AI governance boards in federal agencies. Simultaneously, private entities weigh heavily on economic prosperity, focusing on efficiency and productivity.
The academic sphere and non-governmental entities regularly provide guidance beneficial to public sector organizations. Recently, the World Economic Forum’s AI Governance Alliance published the Presidio AI Framework, providing a step-by-step approach for safe development, deployment, and use of generative AI.
Policies are only the Beginning
While policies provide a roadmap to effective AI governance, the crux lies in their implementation. For instance, despite New York City publishing its own AI Action plan, the AI-powered chatbot set up to aid entrepreneurs misguided users into possibly breaking the law.
Operationalizing governance demands a people-first, accountable approach. Agencies must designate accountable leaders and support them financially, provide governance training, and evaluate inventory beyond initial impact assessments.
Leadership and Training in Governance
The importance of accountable leadership cannot be overstated. Leaders who understand the complex dynamics of data and its inherent biases must be chosen. They must be financial capacitated and held accountable to verify ethically-operated AI that resonates with community values.
Applied governance training is another crucial aspect. Hackathons and innovative tech events can be extended to include AI governance. The agencies establishing policy can judge these events, and provide applied training in creating governance artifacts and assessing ethics and potential risks.
Inventory Evaluation: More than Just Impact Assessments
While algorithmic impact assessment forms are commonly used, there are significant concerns when used in the absence of rigorous training and communication. Incentives for completing these forms, understanding of AI definitions, potential bias, and responsibility for risk are significant factors that need to be addressed.
The need for applied training and a strong organizational culture that consistently measures model behavior against ethical guidelines is urgent and undeniable.
Building a Participatory Culture
Creating a collaborative culture is vital. A diverse, multidisciplinary CoE can enhance employee understanding of AI risks and their impact. Agencies must focus on making governance a part of collective innovation efforts and fostering an environment where responsibility is shared.
To prepare for an AI-driven future, contact HAL149, a leading company that develops AI assistants for organizations, enabling automation and growth while aligning human efforts with AI capabilities.
