美國、中國和歐盟同意在英國峰會上共同應對人工智慧風險

At a summit in the United Kingdom, the United States, China, and the European Union agreed to collaborate on managing the risks associated with artificial intelligence (AI). This rapidly evolving technology has been a cause for concern among tech executives and political leaders, who warn that uncontrolled AI development could pose a significant threat to the world. This has prompted a race among governments and international institutions to establish safeguards and regulations. In a first for Western efforts to ensure safe AI development, a Chinese vice-minister joined US and EU leaders and tech giants like Elon Musk and Sam Altman at Bletchley Park, the site of Britain's World War II codebreakers. Over 25 countries, including the US, China, and the EU, signed the 'Bletchley Declaration', asserting the need for a collective approach to AI oversight.

The declaration outlined a two-pronged agenda: identifying shared concerns and enhancing scientific understanding of them, and developing cross-country policies to mitigate these risks. Wu Zhaohui, China's vice-minister of science and technology, stated at the summit's opening session that Beijing was prepared to enhance collaboration on AI safety to help establish an international 'governance framework'. He emphasized that all countries, regardless of their size and scale, have equal rights to develop and use AI. Concerns about AI's potential impact on economies and society escalated in November last year when Microsoft-backed OpenAI made ChatGPT publicly available. This tool, which uses natural language processing to generate human-like dialogue, has fueled fears among some AI pioneers that machines could eventually surpass human intelligence, leading to unforeseen consequences.

Governments and officials are now striving to chart a path forward in collaboration with AI companies, which fear being burdened by regulation before the technology reaches its full potential. Elon Musk suggested that understanding should precede oversight and proposed the use of a 'third-party referee' to alert when risks arise. While the EU's AI oversight has focused on data privacy and surveillance and their potential impact on human rights, the British summit is examining existential risks from highly capable general-purpose models known as 'frontier AI'. Mustafa Suleyman, co-founder of Google Deepmind, expressed his belief that current AI frontier models do not pose any 'significant catastrophic harms', but advocated for proactive planning as the industry continues to train larger models.