UN Maintenance of international peace and security Meeting

Today´s discussion

Maintenance of International Peace and Security

FROM THE UNITED NATIONS

On December 19th, 2024, the United Nations Security Council convened under the agenda item “Maintenance of International Peace and Security” for an important discussion on AI and its global impact.

Chaired by U.S. Secretary of State Antony Blinken, the meeting featured briefings from UN Secretary-General, António Guterres; Chief AI Scientist at Meta, Yann LeCun; and Co-Director of Stanford University’s Human Centered AI Institute, Fei-Fei Li.

The UN Secretary-General highlighted the risks of inequality related to AI technologies. He urged action to ensure AI benefits everyone, not just a select few, saying AI must never stand for ‘Advancing Inequality.’

Focus on AI Experts : Fei-Fei Li and Yann LeCun

This meeting caught my curiosity thanks to the participation of 2 main gurus of AI : Fei-Fei Li and Yann LeCun.

I really was interested in listening to what such experts had to say about international peace and security.

Who is Fei-Fei Li ?

Fei-Fei Li, often referred to as the “AI Godmother,” is a renowned computer scientist specializing in artificial intelligence (AI) and computer vision. She co-founded the non-profit AI4ALL, which aims to increase diversity and inclusion in AI. As a professor at Stanford University, she has made significant contributions to AI research, particularly in visual recognition technology. Her startup, World Labs, is revolutionizing AI by giving machines 3D spatial intelligence. Fei-Fei is also a leading advocate for ethical AI, emphasizing the importance of human-centered design and responsible development.

Who is Yann LeCun ?

Yann LeCun is a pioneering computer scientist, often called the “Godfather of Deep Learning.” He is the Silver Professor at the Courant Institute of Mathematical Sciences at New York University and serves as Vice President and Chief AI Scientist at Meta. He is widely known for his work in optical character recognition and computer vision through convolutional neural networks (CNNs), a cornerstone of modern artificial intelligence. His contributions earned him the 2018 Turing Award alongside Geoffrey Hinton and Yoshua Bengio. LeCun envisions developing AI systems capable of unsupervised learning, pushing the boundaries of autonomous intelligence.

What Did Yann LeCun Say at the UN Meeting ?

Yann LeCun emphasized that AI will profoundly transform the world, enhancing human intelligence and addressing global challenges like climate change and disease. He stressed the importance of open-source, collaborative AI development, citing Meta's Llama model, which has been downloaded 6.5 million times and spawned 85,000 open models. LeCun highlighted AI's current limitations, such as its focus on text and lack of real-world understanding, but predicted significant advancements in the next decade. He called for international cooperation to ensure accessibility and safety of AI, suggesting initiatives like connecting cultural materials and providing AI supercomputers globally.

yann lecun Action Items

Establish a global network of infrastructure to support the development of AI, enabling people all over the world to participate in the creation of a common resource.
Ensure that the digital diet of citizens is not controlled by a handful of companies.
Perform rigorous testing and red teaming on foundation models to ensure their safe deployment.
Focus international cooperation on connecting cultural material, providing AI-focused supercomputers in multiple regions, and establishing a modus operandi for the distributed training of a free and open, universal foundation model.
Unify the regulatory landscape to ensure the development and deployment of open-source models is not hindered.

AI's Transformative Impact on Society

Yann LeCun emphasizes that AI will profoundly transform the world, amplifying human intelligence and accelerating progress in science and medicine.
AI will facilitate access to information and knowledge, making more people informed, creative, and productive.
In the coming decade, AI will become pervasive, with everyone having access to virtual AI assistants through wearable devices like smart glasses and smartphones.
AI will supercharge productivity and human creativity, helping countries with aging populations and solving major problems like climate change and diseases.

Foundation Models and Open Source AI

AI systems are produced in two phases: training a foundation model and fine-tuning it for specific applications.
Foundation models must be trained on all cultural materials in all languages to be accessible and useful globally.
Two conditions are necessary: foundation models must be free and open source, and training must be collaborative and distributed across multiple data centers.
Governments and the private sector must work together to ensure a global network of infrastructure to support AI development.

Meta's Role in Open Source AI

Yann LeCun highlights Meta's leading role in producing and distributing free and open-source foundation models like Llama.
Llama has been downloaded 6.5 million times since mid-2023, with 85,000 open models derived from it.
Open small models have been used in various sectors, including agriculture, education, healthcare, manufacturing, and environmental preservation.
The open-source nature of Llama has accelerated progress through contributions from academic labs, independent developers, startups, and large corporations.

Safety and Security of AI Applications

AI applications must be deployed safely, undergoing rigorous testing and red teaming.
Historically, open-source platforms have been more secure than proprietary ones.
There is no evidence that current AI presents an existential risk or significantly increased threats over traditional technology.
Current AI systems are focused on text and language, not the real world, and have limitations in understanding, memory, and reasoning.

Future of AI and Its Societal Impact

Yann LeCun predicts that future AI systems will match and surpass human intellectual capabilities within the next decade or two.
These super-intelligent systems will understand the physical world, remember, reason, plan, and have some level of common sense.
These systems will be under control and accomplish tasks given to them, subject to safety guardrails.
AI may bring about a new renaissance, akin to the invention of the printing press, by amplifying human intelligence and disseminating knowledge.

International Cooperation and AI Development

AI can contribute to international peace and security by supercharging the diffusion of knowledge and powering global economic growth.
International cooperation should focus on connecting cultural materials and providing AI-focused supercomputers in multiple regions.
Establishing a modus operandi for the distributed training of a free and open universal foundation model is crucial.
Unifying the regulatory landscape to support open-source model development is necessary to avoid hindrances.

From Yann LeCun Speech, how can we ensure AI models are accessible and beneficial to all communities globally?

There are a few key ways to ensure AI models are accessible and beneficial globally making AI foundation models free, open-source and developed through global collaboration and infrastructure is key to ensuring AI is accessible and beneficial across all communities worldwide, as describes below :

Foundation models must be free and open-source. Mr LeCun states that "foundation models must be free and open source" for them to be accessible and useful to everyone around the world.
Training of foundation models should be done in a collaborative, distributed fashion across multiple data centers globally. The speaker says this is necessary "if we want them to be accessible and useful to everyone around the world."
Governments and the private sector need to work together to establish the global network of infrastructure to support the development of these free, open-source AI foundation models. This will "enable people all over the world to participate in the creation of a common resource."
Unifying the regulatory landscape is also important to ensure the development and deployment of open-source AI models is not hindered. This will help facilitate the global accessibility and adoption of these AI systems.

What Did Fei-Fei Li Say at the UN Meeting ?

Fei-Fei Li, an expert in AI with a focus on computer vision, deep learning, and healthcare, emphasizes the dual potential of AI to both benefit and harm humanity. They advocate for a human-centered AI approach, urging governments to invest in AI research and infrastructure to ensure equitable access and responsible development. They propose a multilateral AI research institute to set global norms and address the digital divide. Emphasizing the need for evidence-based governance, they call for collaboration to overcome AI-related risks like bias and disinformation. The speaker urges public sector leadership, global collaboration, and targeted policy to unlock AI's potential while safeguarding its responsible use.

Fei-Fei Li Action Items

Foster public sector leadership in AI.
Champion global collaboration through a multilateral AI research institute.
Address the digital divide and ensure equitable access to AI globally.
Advance evidence-based governance and a global research agenda for AI.

AI's Transformative Potential and Global Responsibility

Fei-Fei Li emphasizes the extraordinary intersection of scientific possibilities and global responsibility in AI, highlighting their 25 years of experience in computer vision, deep learning, robotics, and AI for healthcare.
She discusses the duality of AI, noting its potential to both help and harm, and stresses the importance of public sector leadership and a human-centered AI approach.
Examples of AI's positive impact include robots navigating disaster zones, precision agriculture systems, and advanced medical imaging tools, which demonstrate AI's potential to drive scientific discovery and improve the world.
She calls for vigilance to prevent the misuse of AI capabilities, underscoring the need for a balanced approach to AI development and deployment.

Ensuring Broad Access to AI Benefits

Fei-Fei Li advocates for a vibrant and healthy AI ecosystem to ensure that AI's benefits are widely distributed and aligned with public interests.
She highlights the concentration of AI innovation in large corporations and select nations and calls for broader access to AI resources.
Governments are urged to adopt a "moonshot mentality" for AI, committing to visionary public sector investment beyond incremental reforms.
The United States' national AI research resource pilot program is mentioned as a step towards democratizing access to compute resources and government data sets, but she calls for global scaling of such efforts.

Global Collaboration and AI Governance

Fei-Fei Li emphasizes the importance of global collaboration in effective AI governance, sharing their experience on the United Nations Secretary General Scientific Advisory Board.
She calls for a multilateral AI research institute, Miri, to bring together experts across disciplines and pull resources across nations, advancing technical innovation and setting global norms for responsible AI development.
Global collaboration is essential to address the digital divide and ensure equitable access to AI tools, training, and infrastructure, benefiting both wealthy and marginalized regions.
She stresses that the benefits of AI should not be reserved for wealthy nations, highlighting the collective responsibility to ensure global stability through equitable AI access.

Evidence-Based Governance and Future Generations

Fei-Fei Li discusses the complexity of AI and the need for careful, evidence-based governance to pave the way for innovation and global progress.
Identified risks such as algorithmic bias, disinformation, and misuse of autonomous systems are acknowledged, with a call for collaboration and ingenuity to overcome these hurdles.
A global research agenda is proposed to fill knowledge gaps and inform targeted interventions, bridging the gap between developers and policy makers.
She reflects on their role as a researcher, teacher, and mother, urging vigilance, collaboration, and a shared commitment to human dignity and global stability in realizing AI's potential.

Call to Action for Public Sector Leadership and Global Collaboration

Fei-Fei Li calls for urgent and united action to foster public sector leadership, champion global collaboration, and advance evidence-based policy.
She emphasizes the need for sustained public investment to ensure AI reflects diverse needs and values, fostering an environment where AI is safe, equitable, and a source of inspiration and progress.
The speaker concludes by reiterating the importance of acting with urgency and unity to unlock AI's transformative potential while safeguarding its responsible development.

From Mrs Li Speech, what steps can be taken to establish a multilateral AI research institute to promote global collaboration and responsible innovation?

From her speech, the key steps that can be taken to establish a multilateral AI research institute to promote global collaboration and responsible innovation. By taking these steps, the multilateral AI research institute can become a hub for responsible innovation, driving technical advancements while upholding democratic values and ensuring equitable access to the benefits of AI:

Secure Funding and Partnerships: Engage with governments, international organizations, and private sector partners to secure the necessary funding and resources to establish the institute. Leverage public-private partnerships to pool resources and expertise.
Develop a Global Governance Framework: Work with stakeholders to create a robust governance framework for the institute that ensures democratic values, international peace and security, and equitable access are at the core of its mission and operations.
Identify Participating Nations and Institutions: Reach out to leading AI research hubs and institutions around the world to participate in the institute. Aim for broad geographic representation to foster global collaboration.
Define the Institute's Mandate and Research Agenda: Establish the institute's mandate to focus on advancing technical innovation, setting global norms for responsible AI development, and informing evidence-based policymaking. Develop a comprehensive research agenda that addresses key challenges and opportunities.
Establish Inclusive Governance and Decision-Making: Ensure the institute has an inclusive governance structure that gives voice to diverse stakeholders, including researchers, policymakers, civil society, and underrepresented communities.
Promote Knowledge Sharing and Capacity Building: Facilitate the exchange of knowledge, best practices, and resources among participating institutions. Invest in training and education programs to build AI expertise globally.
Advocate for Supportive Policies and Funding: Engage with policymakers to advocate for sustained public investment in the institute and supportive policies that enable its success, such as incentives for global collaboration.

Where to Watch the Replay ?

You can watch the whole meeting on the United Nations web TV channel.

Want to know more about Safe AI and Responsible AI ? Just contact us.