Our Founding Symposium took place on 9 September 2024 at the artience headquarters in Kyobashi Edogrand, Tokyo, with about 80 participants in person and about 50 online. This second half of the report covers Part 2, “Industry–Government–Academia Collaboration toward a Hopeful Future for AI and Humanity.” Part 1 covers the first half. Written by Mirei Takahashi.
- Toshikazu Fukushima (Fellow, Japan Science and Technology Agency), “Research Trends in Next-Generation AI and AI Risk Countermeasures.” He described two currents, next-generation AI models and measures against AI risk, and argued that reducing risk requires not only dealing with individual problems but also mechanism design and trust building in a multi-agent society.
- Masahito Iida (Researcher, Mitsubishi Research Institute), “An Overview of AI Governance.” Drawing on his institute’s surveys, he reviewed how Japanese companies use generative AI and the risks they see, and compared AI governance in Europe, the United States, China and Japan, including the Hiroshima AI Process.
- Akiko Murakami (Executive Director, Japan AI Safety Institute), recorded address on domestic collaboration in AI safety. She explained the role of the institute, established in February 2024, in supporting government and the private sector, working with other AI safety institutes on international standards and building a community, and called for this to extend to private organizations, including ALIGN.
- Panel, “Toward an Industry–Government–Academia Network for a Hopeful Future for AI and Humanity,” moderated by our co-founder Ryota Kanai, with Akiko Aizawa (Professor, National Institute of Informatics), Fumiko Kudo (Specially Appointed Associate Professor, Osaka University), Hiroaki Sakuma (Secretary General, AI Governance Association) and our chair Koichi Takahashi. Aizawa said such a network should take on public-interest problems that AI cannot solve; Kudo discussed why long-term risks are hard to address in policymaking; Sakuma described the work of the AI Governance Association, with about 70 member companies, and its wish to work with ALIGN on the underlying research.
- Takahashi’s summary of the panel. He made three points: alignment is a guardrail rather than a brake for the AI industry; long-term risks should be turned into solvable problems; and AI alignment needs to grow as a highly interdisciplinary field.