WEF papers propose framework for responsible AI to mitigate risks

A global effort is needed to create equitable access to artificial intelligence (AI) which holds the potential to address challenges but also poses risks of widening existing digital divides, a series of new papers released at the World Economic Forum (WEF) at Davos said. 

Three new briefing papers by the WEF’s AI Governance Alliance offer recommendations on building safe systems and technologies, ensuring responsible applications and transformation, and advancing resilient governance and regulation.

The papers focus on generative AI governance, unlocking its value, and a framework for responsible AI development and deployment.

On international cooperation and inclusive access in AI development and deployment, Generative AI Governance: Shaping Our Collective Global Future evaluates national approaches, addresses key debates on generative AI, and advocates for international coordination and standards to prevent fragmentation.

The paper says that equitable access and inclusion of the Global South in all stages of AI development, deployment, and governance is critical for innovation and for realizing the technology’s socioeconomic benefits and mitigating harms globally.

“Access to AI innovations can empower jurisdictions to make progress on economic growth and development goals. Genuine access relies on overcoming structural inequalities that lead to power imbalances for the Global South, including in infrastructure, data, talent, and governance,” the paper argues.

“To adequately address unique regional concerns and prevent a relegation of developing economies to mere endpoints in the AI value chain, there must be a reimagining of roles that ensure Global South actors can engage in AI innovation and governance,” the paper published in collaboration with Accenture further says.

It recommends international coordination, compatible standards, and flexible regulatory mechanisms to prevent a fracturing of the global AI governance environment into non-interoperable spheres with prohibitive complexity and compliance costs. 

Commenting on the company’s partnership with the alliance, Accenture’s chief technology innovation officer Paul Daugherty said, “The evolution of AI is unique in that the technology, regulation, and business adoption are all accelerating exponentially at the same time.”

“It’s critical that the public and private sector come together to share insights, resources, and best practices for building and scaling AI responsibly. Leaders in this space must prioritize inclusive AI so that the benefits of this technology are shared in all parts of the world, including emerging markets,” Daugherty added. 

The second paper—titled Unlocking Value from Generative AI: Guidance for Responsible Transformation—provides guidance on the responsible adoption of generative AI, emphasizing use case-based evaluation, multistakeholder governance, transparent communication, operational structures, and value-based change management for scalable and responsible integration into organizations.

The briefing paper argues that organizations should evaluate potential use cases across three domains: business impact, organizational readiness, and investment strategy.

The report says that a multistakeholder approach helps leaders to mitigate risks such as impact on the workforce, sustainability, or hallucinations.

Value-based change management is critical to addressing human impact and ensuring the workforce remains engaged and upskilled, the paper, written in collaboration with IBM Consulting, says. 

The third briefing paper—Presidio AI Framework: Towards Safe Generative AI Models—addresses the need for standardized perspectives on the model lifecycle by creating a framework for shared responsibility and proactive risk management of AI.

The Presidio AI Framework consists of three core components: a comprehensive end-to-end view of the generative AI life cycle; robust guardrails to be considered at different steps of the generative AI life cycle; and shift-left methodology proposing the implementation of guardrails at the earliest stage possible in the generative AI life cycle. 

The paper, prepared in collaboration with IBM Consulting, also emphasizes the need for greater multistakeholder collaboration between industry stakeholders, policymakers, and organizations.

IBM vice-chairman Gary Cohn said the company continues to drive responsible AI and governance. 

“We all have an obligation to collaborate globally across corporations, governments, and civil society to create ethical guardrails and policy frameworks that will inform how generative AI is designed and deployed,” Cohn said. 

AI Governance Alliance brings together governments, businesses, and experts to shape responsible AI development applications and governance, and to ensure equitable distribution and enhanced access to this path-departing technology worldwide.

AIGA also seeks to mobilize resources for exploring AI benefits in key sectors, including healthcare and education.

Commenting on the initiative, Omar Sultan Al Olama, the United Arab Emirates (UAE) minister for AI, digital economy, and remote work applications, said, “This roadmap is a testament to our belief in AI as a tool for universal progress and equality, and it embodies our commitment to a future where technology serves humanity in its entirety.” 

Paula Ingabire, minister of information communication technology and innovation of Rwanda, said the country’s inclusion in the AI Governance Alliance aims to ensure that the region does not play catch up in shaping the future of AI governance and accessibility.

The minister also announced a high-level summit on AI in Africa towards the end of 2024, to be held in collaboration with the World Economic Forum.

“The event’s primary goal will be to align African countries on common risks, barriers, and opportunities and, ultimately, devise a unified strategy for AI in Africa,” she said.