Global leaders and top executives from the artificial intelligence (AI) industry gathered in Seoul this week for an AI safety summit. Co-hosted by the UK and South Korea, the summit aimed to establish basic safety measures for the rapidly advancing field of AI. This meeting follows the inaugural global summit on AI safety held at Bletchley Park in England six months ago.
During the summit, the British government announced the establishment of an international network dedicated to accelerating the advancement of AI safety science. Ten countries, including Australia, Canada, France, Germany, Italy, Japan, Singapore, South Korea, the UK, the US, and the European Union, have signed this agreement.
The first day of the summit saw world leaders and top AI companies participating in a virtual meeting. Discussions, chaired by UK Prime Minister Rishi Sunak and South Korean President Yoon Suk Yeol, revolved around AI safety, innovation, and inclusion. The leaders agreed to the Seoul Declaration, emphasizing “human-centric, trustworthy, and responsible” international collaboration in building AI to address major global issues.
Sunak highlighted the importance of AI safety in a statement, expressing enthusiasm for AI technology and the UK’s role in its global development. He emphasized the essential nature of ensuring safety in AI and praised the agreement for a network of AI Safety Institutes as a significant step toward managing potential risks associated with AI technology.
Additionally, the summit saw 16 AI companies, including Amazon, Google, IBM, Microsoft, and Samsung Electronics, make safety commitments. These companies agreed not to develop or deploy any model or system if mitigations cannot keep risks below certain thresholds, marking a significant global agreement on AI safety. The summit also acknowledged the risks posed by rapidly evolving AI technology, including existential threats to humanity, AI inequality, data scarcity, copyright issues, and environmental impact due to high electricity consumption by AI data centers.