Current Global Cooperation: Ensuring Treaty Compliance and Further Advancements
In our rapidly evolving world, the ethical usage of artificial intelligence (AI) is a pressing concern. At a recent panel in Davos, experts discussed the approaching era of artificial general intelligence (AGI), where AI surpasses human capability in a wide range of tasks.
Three out of four panelists agreed that AGI is on the horizon within the next 5 to 10 years, with only one suggesting a 20-year timeline. The consensus was that we are rapidly approaching a stage in AI development where it will create its own theories, much like humankind, or develop new games like "Go."
The panelists discussed the international implications of this potential future. One of the concerns raised was the apparent divide between the United States' "win the AI war" mentality and a more collaborative approach. The consensus among the panelists was that cooperation would be the ideal way to approach this new technology, benefiting all humanity.
Panelist Berkeley professor Dawn Song emphasized the need for international collaboration in developing AI safety measures. She highlighted that our understanding of AI is limited, as it has capabilities that even developers struggle to predict. Thus, she argued that it is crucial to develop guidelines and guardrails to control the development of agentic AI, which are systems equipped with goals and self-preservation mechanisms.
The panel also addressed the potential consequences of non-agentic AI outcompeting those with human-in-the-loop (HITL) systems. Panelist Yoshua Bengio from Université de Montréal warned that entering the era of agentic AI is a significant threshold moment, making it even more important to ensure that these systems are safe.
Overall, the panel highlighted the importance of international cooperation in ensuring the safe and responsible development of AGI. They discussed the need for institutions like a "CERN for AI" and a supervening body similar to the United Nations to help coordinate efforts. They also called for more research and dialogue to increase awareness and understanding of AGI risks to inform international treaties.
- Given the approaching era of AGI and its potential impact, there's a growing need for policy discussions at the national level regarding the deployment of finance and VC in this field, ensuring ethical and secure development.
- The panel at Davos also suggested the establishment of a global institution akin to CERN for AI, and a supervising body similar to the UN, to manage the timelines of AGI development and promote international collaboration in superintelligence research.
- As AGI approaches, it's crucial to consider the potential risks, such as non-agentic AI outpacing HITL systems, and establish guidelines and policies to minimize these dangers, advocated by experts like Dawn Song and Yoshua Bengio.
