This is a grand event for artificial intelligence, a collective step towards a better future.
On July 6, the three-day 2024 World AI Conference&High-Level Meeting on Global AI Governance (WAIC 2024) concluded. With the theme “Governing AI for Good and for All,” a total of 107 forums were held to discuss the mathematical limits of scaling laws, explored how AI empowers scientific research and industrial development.
The closing marks the beginning of a new journey as the exploration of AI has no end.
Promoting “AI for Good” Through Global Consensus
During the three-day event, 1,300 global leaders from over 50 countries and regions gathered together, including nine recipients of the Turing Award, Fields Medal, and Nobel Prize, as well as 88 top academicians from around the world. Like stars converging, the seeds of inspiration they planted are sure to take root and grow in the near future, becoming a powerful force driving technological advancement and civilizational progress.
Unlike previous years, this year’s conference brought the governance of AI to an unprecedented level.
The rapid development of AI technology has brought tremendous benefits to economic and social development, but it has also introduced numerous risks and challenges, such as deepfakes, privacy protection, and intellectual property disputes. Promoting the development and safety of artificial intelligence, as well as addressing global governance and ethical issues, are major unavoidable topics that urgently require consensus through dialogue and cooperation.
Xue Lan, a senior professor at Tsinghua University and Dean of Schwarzman College, stated that the
the World Artificial Intelligence Conference,themed “Governing AI for Good and for All,” captures the latest developments in AI development and governance. From the perspective of global AI governance, issues and challenges are primarily concentrated in areas such as geopolitics, technological pathways, the coordination of governance and technological progress, mechanism construction, and supervision and enforcement. Therefore, it is essential to strengthen collaborative governance, expand multi-level and multi-channel bilateral and multilateral exchanges and dialogues, and promote international cooperation in the field of AI for major scientific endeavors to address common global challenges. “We need to push for the AI governance system to become an international public good, rather than a small, exclusive club.”
Indeed, AI governance is not a challenge that any single country can face alone. Amandeep Singh Gill, the UN Secretary-General’s Envoy on Technology, stated at the conference’s opening that AI is extremely important, and relying solely on market forces, dispersed national efforts, and multilateral forces carries too much risk. If we want to fully reap the benefits of AI and achieve sustainable development goals, international cooperation is essential.
Echoing this theme, during this year’s conference, ministerial-level officials from over 30 countries and leaders of relevant international organizations held a significant international dialogue specifically focused on AI governance. The participants exchanged in-depth views on issues related to AI development, safety, and governance, expressing their willingness to actively engage in exchanges and cooperation. They aim to collaboratively build an open, fair, and effective governance mechanism to jointly promote the healthy, safe, and orderly development of AI.
Many international experts have expressed their views on global AI governance. Gaël Varoquaux, research director at French Institute for Research in Computer Science and Automation (Inria) stated, “We believe there needs to be international global governance because both the internet and AI can easily cross borders.” Sassine Ghazi, president and chief executive officer of Synopsys, emphasized that global collaboration can enable people around the world to benefit from the significant advancements brought by AI while ensuring that innovation is practiced responsibly.
The participating experts also stated that global AI governance cannot be separated from the support of cutting-edge technologies. Regarding AI safety governance, Turing Award winner Chi-Chih Yao believes that AI safety governance should not be a patchwork, addressing one issue at a time as it arises. Instead, it is necessary to explore beneficial general artificial intelligence that aligns AI with human interests. During the design phase, mathematical principles should be incorporated to ensure that all decisions made by robots are human-centered. Robots must communicate with humans to understand their needs. The goal is to design mathematically absolutely safe general artificial intelligence, where humans do not interact directly with large models but only with provably safe “white boxes.”
Zhou Bowen, Director and Chief Scientist of the Shanghai Artificial Intelligence Laboratory, proposed the exploration of the “AI-45° Law.” He acknowledged that AI development is currently imbalanced, with AI model safety lagging far behind performance. This imbalance is due to significant differences in talent density, commercial drive, and computing power investment between safety and performance. In the long term, AI should generally develop along a 45° balance between safety and performance. There can be short-term fluctuations, but it should not remain significantly below or above 45° for extended periods, as this would hinder technological and industrial application development. There are many technical paths to achieve the 45° balance law in AI, and the Shanghai Artificial Intelligence Laboratory is exploring a path centered on causality, referred to as the “causal ladder” of trustworthy AGI.
Accelerating“AI Empowerment” through Frontier Exploration
Development and safety are the two wings of AI. Based on safety and governance, exploring the forefront of AI and promoting better empowerment for industrial development is also a key topic of this year’s conference.
AI is the wisdom engine of scientific exploration and the catalyst driving industrial development. In early 2023, a cover article in Nature presented a pessimistic view of the current state of research paper development, stating that “scientific progress is slowing down.”
However, it is comforting that AI is reshaping a new paradigm of scientific discovery. As experts discussed in the forums, large models compress world knowledge internally, and the uncertain generative properties may help humanity break through knowledge silos across different disciplinary domains to achieve innovative exploration. For instance, astronomy, as a discipline with massive data, represents one of the best scenarios for AI application and driving scientific discovery. Artificial intelligence, with its extraordinary data processing capabilities, unveils the mysteries of the universe’s depths.
AI for Science is also driving a new paradigm shift in scientific research. Scaling laws have transformed AI, but relying solely on compressing and summarizing massive amounts of data is not enough to achieve general artificial intelligence. Qi Yuan, Director of the Shanghai Institute of Science Intelligence, pointed out that, much like AlphaGo’s victory over Lee Sedol, humanity has seen the potential for machines to surpass human cognition. However, more importantly, AI should strive to have self-learning and innovative capabilities, akin to AlphaZero. “The ultimate manifestation of AI for Science is to create an ‘AI Einstein’ that can understand complex worlds and discover unknown laws.”
From exploring frontier technology to empowering industrial development, AI is penetrating various industries at an unprecedented pace. It deeply integrates into the entire production and operation process, reducing production costs, enhancing production efficiency, boosting core competitiveness, fostering new business models, and reshaping traditional industries.
Experts indicate that in the era of intelligence, AI extensively empowers human beings, tools, resources, and technological factors of production. The “surge” in the generalization ability of large models accelerates knowledge discovery, expands human capability boundaries, and continues to upgrade industrial digitization and intelligence, thereby driving transformation in production paradigms.
Industry professionals deeply feel that AI is a crucial engine for new-quality productivity, serving as the “productivity of productivity.” AI is optimizing workflows for video creation, becoming a new engine for innovation-driven advancements. The millisecond-level code generation speed of AI continuously enhances developers’ coding efficiency. AI-driven autonomous driving is becoming a reality, transforming the future of transportation. The application scenarios of large models in industry are becoming increasingly diverse, penetrating into research and development, design, manufacturing, and operation management, ushering in a new era of intelligent manufacturing. From question-and-answer tools to assistive tools, and now intelligent agents, large models are evolving from general scenarios to industry-specific applications, demonstrating increasingly significant productivity improvements.
Embodied intelligence based on humanoid robots is currently one of the frontier interdisciplinary hotspots in academia and industry. It represents an important developmental direction towards general artificial intelligence.
Building an “AI Ecosystem” Through Continuous Efforts
The gathering of elite scholars and industry leaders from around the world made the conference shine brightly.
The accompanying exhibitions allowed attendees to directly experience the charm of AI. Over 500 well-known companies showcased over 1,500 exhibits, with more than 50 new products debuting, marking historical highs. Nearly a hundred large models were prominently displayed, alongside 18 humanoid robots forming a matrix, and the release of China’s first open-source full-scale general humanoid robot. These innovations brought about by AI are truly eye-opening.
Talent is the primary resource for AI development. Focused on building a talent ecosystem, the SAIL Award (Superior AI Leader Award), the WAICYOP Award(World Artificial Intelligence Conference Youth Outstanding Paper Award), WAIC Yunfan Award, the Fourth Best Practices of Applied Algorithms, the Youth AI Innovation Competition, and the Puyuan Large Model Challenge have become integral parts of the World Artificial Intelligence Conference. “Three Awards ,Three Competitions” not only represent authoritative recognition of outstanding achievements in the field of AI but also harness powerful momentum to drive global advancements in AI technology. They empower more young people to break new ground in the journey of AI.
The World Artificial Intelligence Conference serves as a window to observe global AI development and is also a microcosm of building the AI ecosystem. Governments, enterprises, universities, research institutions, and capital collaborate to construct a favorable ecosystem characterized by resource sharing, complementary advantages, and collaborative innovation. This ecosystem synergy promotes rapid technological iteration and industrial upgrading, attracting global attention to the field of AI.
Although WAIC 2024 has concluded, it marks the beginning of a new journey. Next year, the 2025 World Artificial Intelligence Conference will be held as scheduled, bringing together friends from around the world to explore the boundless possibilities of AI together.
Media Contact
Company Name: Donghao Lansheng (Group) Co. Ltd.
Contact Person: Sharon Gong
Email: Send Email
City: Shanghai
Country: China
Website: www.worldaic.com.cn