Summary
- UN Secretary-General António Guterres inaugurated the first Global Dialogue on AI Governance in Geneva, emphasizing that AI's growth is outpacing the ability of its creators to manage it.
- He used the term "vibe coding," a metaphor for insufficient oversight in AI development, stating, "We cannot vibe code the future of humanity."
- Guterres called for a global prohibition on lethal autonomous weapons and introduced an AI Child Safety Pledge, with a follow-up dialogue planned for New York in 2027.
During the inaugural Global Dialogue on AI Governance in Geneva on Monday, UN Secretary-General António Guterres addressed representatives from 193 nations, warning that the rapid advancement of AI is outpacing the governing bodies designed to oversee it, likening humanity's current situation to conducting an experiment on itself "without a plan, and without consent."
"Artificial intelligence is advancing at runaway speed,” Guterres remarked in his keynote address. He noted that this technology has the potential to reshape economies, influence elections, and alter security dynamics, yet it is being deployed more quickly than even its creators can manage.
The event, which gathered all 193 UN member states, represented a multilateral effort to establish governance over a rapidly evolving technology.
Guterres illustrated the risks associated with unchecked AI by referencing "vibe coding," a concept introduced by Andrej Karpathy, a founding member of OpenAI and former Tesla AI director. This term describes a programming approach where users express their desires to AI without closely monitoring the process. It has recently been added to the Merriam-Webster dictionary.
He acknowledged that while vibe coding "can do wonders" and is increasingly trusted, "we cannot vibe-code the truth,” asserting the necessity of careful oversight in shaping humanity's future.
Guterres highlighted that while the internet took 15 years to connect a billion users, AI achieved this milestone in just two years. He emphasized that current AI systems have evolved beyond mere tools, as they are now capable of writing code, acting autonomously online, and making decisions with decreasing human input.
"Our institutions were designed to manage machines that follow commands. They are not equipped for machines that make decisions," he added.
The dialogue is a product of the 2024 Global Digital Compact, which for the first time mandates international governance of AI under the UN. The opening session also presented a preliminary report from the Independent International Scientific Panel on Artificial Intelligence, comprising 40 scientists from 140 countries, which concluded last week that there is currently no guarantee AI won't lead to catastrophic harm.
Guterres cited three primary concerns identified by the panel: speed, power concentration, and the erosion of truth. The second concern relates to the concentration of computing resources, data, and talent among a few companies and nations, leaving most of the world excluded from crucial decisions shaping AI development. This is a worry shared by other AI experts, including Yann Lecun, Andy Kowinski, and Yoshua Bengio.
The third concern involves the potential for AI to generate misinformation that can manipulate public perception just as effectively as verified facts, undermining what Guterres termed "the integrity of our information ecosystem."
As part of his proposals, Guterres introduced an AI Child Safety Pledge, which would require companies to validate the safety of AI products for children through independent testing. He advocated for a zero-tolerance policy regarding the generation of child sexual abuse material and emphasized the need for real human support for distressed children, rather than relying on chatbots. "No child should be a guinea pig for unregulated AI," he stated.
Additionally, he condemned lethal autonomous weapons—machines capable of selecting and killing targets without human intervention—as "morally repugnant" and called for an international ban on their use.
Guterres encouraged states to engage in discussions without delay. The next session of the Dialogue is set to take place in New York in 2027. He also urged the General Assembly to establish a Global Fund for AI to enhance computing access in developing nations and challenged major AI companies to ensure all data centers operate on renewable energy by 2030, the same year he predicts these facilities will surpass all but five nations in electricity consumption.