Special Envoy: Global AI Summit Aims to Bridge Digital Divide and Promote 'AI Biodiversity'
Speaking at the Paris Peace Forum, Anne Bouverot, Special Envoy for the upcoming AI Action Summit, explained her vision for making artificial intelligence more inclusive and environmentally sustainable.
The summit, scheduled for February 10-11, 2025, in Paris, marks a significant shift in the global conversation around AI, moving from theoretical risks to practical actions.
"We don't want AI to be captured by a few predators," Bouverot emphasized, echoing President Macron's call for "AI biodiversity." This perspective comes at a crucial time, as the world grapples with an emerging AI divide between makers, takers, and the deprived, she added.
The summit has already garnered significant interest, with 773 AI projects submitted from 102 countries, including 232 from Africa and 127 from Asia. The global response underscores the urgent need for broader participation in AI development, moving beyond the current paradigm dominated by a handful of large tech companies, Bouverot said.
One of the summit's key initiatives is the proposed creation of an AI Foundation, designed to support the development of smaller, more focused AI models. This approach contrasts sharply with the current trend toward massive language models that require enormous computing power and energy resources. "We need to have targeted small models to cater to specific needs for specific countries, specific communities, specific industry sectors, specific SDGs," Bouverot explained.
The environmental impact of AI has emerged as a pressing concern. Large AI models are increasingly driving tech companies to seek out energy-rich locations and even consider small nuclear reactors for their data centers. The summit will address these challenges by promoting more efficient, targeted AI solutions while also exploring AI's potential benefits for climate action, such as improving battery technology and predicting extreme weather events.
Inclusivity stands at the heart of the summit's agenda. Rather than allowing AI development to remain concentrated among a few wealthy nations and companies, the initiative aims to enable broader participation. This includes making AI more accessible to smaller enterprises and public administrations in developing countries, which currently struggle to afford AI solutions, Bouverot said.
The summit will bring together an impressive array of stakeholders, including representatives from about 100 countries, hundreds of CEOs, academics, and civil society organizations. Unlike previous AI gatherings that focused primarily on safety and risks, this summit emphasizes concrete actions and practical implementations.
Regarding regulation, Bouverot advocates for a measured approach. While acknowledging the need for oversight, she suggests that it's too early for comprehensive global regulation. Instead, the summit will focus on building partnerships and achieving specific agreements on particular aspects of AI development and deployment.
The initiative represents a marked departure from the way social media was allowed to develop without adequate forethought about its societal impacts. "We let social media develop without really thinking about it, and now a bit too late, probably we're realizing that we should have helped its development in a direction that would be more beneficial to our societies," Bouverot noted.
China's participation in the summit is also anticipated, following a joint declaration on AI made during President Xi's state visit in May. Bouverot plans to visit China to discuss their concrete contributions to the summit, maintaining France's tradition of inclusive international dialogue.
The summit organizers are currently recruiting partners, including countries, philanthropic organizations, and enterprises to support these initiatives. While the February summit will serve as a launching point, the work is expected to continue well beyond, potentially involving multilateral development banks and other international institutions.
"We need to really do everything we can to mitigate the risks, and we need to do everything to help AI develop towards the positive outcomes," Bouverot said. "This is such an important matter for society that we cannot afford to get it wrong."