The UN Is Speaking Up About AI — Here’s What They’re Saying | Ian Bremmer and Bilawal Sidhu | TED

28 Sep 2024 (23 days ago)

AI Governance: The Need for Global Collaboration

António Guterres, the UN Secretary-General, has identified climate change and disruptive technologies, particularly AI, as two global issues where governance is inadequate and require urgent attention. (2m31s)
AI's rapid advancement, coupled with significant investments and its transformative impact on human interaction, innovation, and the global economy, necessitates global governance to harness its potential and mitigate risks. (4m0s)

AI presents opportunities for individuals, enabling them to utilize AI tools for various purposes like creativity, communication, and problem-solving, while also offering advancements in industries such as healthcare, energy, and transportation. (5m1s)
There is a lack of governance surrounding AI, with the private sector, primarily driven by profitability, leading its development and deployment. (6m52s)
The current trajectory of AI development, primarily influenced by market forces, could lead to a "technopol" world where a small number of technologists hold significant power over AI's direction and applications. (10m33s)
People in the AI industry are primarily focused on business models and market agendas, which may lead to neglecting opportunities in the global South, among the poor, and in biodiversity. (20m38s)
AI poses a significant threat to economic systems, with the potential for malicious actors to cause financial market crashes or critical infrastructure failures. (45m1s)
The increasing influence of algorithms on human behavior, particularly through AI, could lead to a loss of humanity as people engage more with non-human actors and become more programmable. (46m48s)

Governments and multilateral organizations need to collaborate with the private sector to establish effective AI governance, as they currently lack the knowledge and expertise to do so independently. (13m14s)
Governments need to collaborate with the private sector on AI to effectively serve the needs of their citizens. (13m54s)
The UN believes that to effectively govern AI, a common understanding of its definition, potential, opportunities, and concerns is necessary, similar to the approach taken with climate change. (14m8s)
The UN's objective in AI governance is not to enforce behavior but to establish a shared understanding of the opportunities and risks associated with AI, which will then enable effective governance. (15m48s)
There is a need for a global conversation to establish common standards and a shared understanding of how AI can be used for the greatest impact, preventing wasted resources and "AI washing." (21m58s)
Global cooperation is needed to address the challenges of AI, similar to how global cooperation is necessary for issues like climate change and financial markets. (50m10s)

Two recommendations are proposed: an AI capacity development network to provide expertise, compute resources, and training data to the global South, and a global fund for AI to support this development capacity. (23m29s)
There is a desire for "meaningful openness" in AI development, allowing individuals, enterprises, and governments to utilize AI effectively. (32m36s)
Two potential governance frameworks for AI are proposed: arms control agreements between nations and a global technology stability board similar to the financial stability board. (33m46s)
The need for a "Ministry of the Future" for AI is suggested, with appointed officials acting independently to address potential threats and ensure responsible development. (38m5s)
The UN is establishing an international scientific panel on AI, similar to the intergovernmental panel on climate change, to define opportunities, risks, and uncertainties related to AI development. (40m47s)
The international community should work together to ensure AI is not used in autonomous weapons systems that make life-or-death decisions. (44m29s)

A small number of corporations and governments will likely be the only entities capable of handling the amount of compute, energy, and water required for AI. (26m11s)
Meta is making a bet on open source AI models, but most of the money in AI is not heading in that direction. (27m11s)
If AI is deployed in real time with access to massive data sets, it might prove to be more efficient than what the private sector can do by itself. (27m55s)
There is a need to determine if AI will be controlled by a few tech companies or if it will be accessible to individuals and nations globally. (50m47s)
The future of AI depends on understanding its complexities, demanding transparency, and promoting responsible development without hindering innovation from startups. (51m31s)