Researchers, policymakers, and behavioral scientists gathered in Hakodate, Japan, for four days in March to discuss whether digital ethics can be more than just a good idea and whether Japan might be the nation that actually develops it.
The future of global AI governance is not a topic that should be discussed in Hakodate. It is located far from Tokyo’s ministries and the Kanto plain’s research clusters at the southernmost point of Hokkaido, Japan’s northernmost main island. A Russian Orthodox church, a morning fish market that predates every digital system anyone in the conference halls was talking about, and Western-style brick buildings are just a few of the remnants of a city that opened to the world before the majority of Japan. The harbor is ancient. However, the questions posed inside Future University Hakodate during the city’s hosting of the 21st International Conference on Persuasive Technology in March 2026 felt, for at least four days, like the right questions being asked in the right place.
The term “Hakodate Consensus,” which has been applied informally to the concepts that have emerged from this meeting, is neither a formal document nor a signed proclamation. It is, if anything, a shared perspective: the conviction that technology should affect human behavior through honest design rather than coercion or manipulation, and that developing the ethical frameworks to uphold that distinction necessitates real cross-border collaboration as opposed to the ethics-washing that scholars have been documenting for years. AI governance frameworks, such as Japan’s own Society 5.0 initiative, run the risk of becoming more aspirational than operational, strong on principles but weak on accountability, as James Wright and others have written extensively about. The scientists in Hakodate were aware of that danger. A number of them had made contributions to the scholarly literature on this particular issue.
Japan’s place in the international discussion of AI ethics is genuinely complex, and it is better to accept that complexity than to try to hide it. According to UNESCO data, the nation spends third in the world on research and development, but the Global Entrepreneurship Monitor ranks it 47th in terms of entrepreneurial intentions. This disparity reflects a genuine aspect of the culture: a consensus-driven, precision-first approach to institutional decision-making that moves slowly when speed is required but ultimately results in something more durable. According to Hitachi’s CTO, it combines AI capabilities with conventional operational technology to create what the company refers to as Practical AI. NEC has developed systems aimed at transferring the knowledge of skilled workers who are getting older to the following generation. When you think about it, persuasive technology research is attempting to incorporate an instinct to enhance rather than replace into digital design more generally, which is reflected in both examples.
Although the tension that the Hakodate gathering attempted to maintain is not new, it is getting more difficult to ignore. In different hands and with different incentives, the same behavioral mechanisms that support someone in maintaining an exercise regimen or managing chronic hypertension can also be used to keep someone doubting an election outcome or scrolling past midnight. The conference’s best paper, written by Kiemute Oyibo, used empirical precision to measure the “darkness” of dark patterns. It was a sort of ethics cartography that mapped the area between beneficial design and overt manipulation. Mapping is an essential task. However, it might only be significant if the frameworks based on it have real regulatory weight rather than merely academic legitimacy.
Watching Japan navigate this specific stage of AI development gives me the impression that the nation’s well-known caution about failure is being reframed as something more beneficial than it first seems, albeit slowly and not without opposition. As part of Forbes Japan’s coverage of the nation’s technological development, Kumiko Seto argued that the aborted H3 rocket launch was an example of a functioning fail-safe rather than a failure in any significant sense. When properly understood, setbacks are data. Similar reasoning appears to be being applied to digital ethics by the Hakodate gathering: the inadequacy of earlier frameworks is not a reason to reject them. It is a case for improved ones that have been developed by more people and put through more rigorous testing under actual circumstances.
It’s genuinely unclear if the ideas that were discussed in those open-plan corridors above the harbor will eventually become policy. Legislation is not produced by academic conferences, but rather proceedings. However, the next meeting takes place in Santiago de Compostela, where the discussion continues. Researchers who traveled there by plane from Qatar, Australia, Canada, and all over Japan worked through four days of papers and arguments before leaving with something that is a little more difficult to identify than a document but no less authentic as a result.
