鈴木健：アルゴリズム、ニュース、そして民主主義の対話

Ken Suzuki founded SmartNews in 2012 and still runs it as CEO. The app has passed 50 million downloads. On April 26 he speaks at the Tech for Impact Summit 2026 in Tokyo about what a news algorithm owes the people who read it.

Most news apps rank stories by what keeps a reader scrolling. SmartNews has spent over a decade trying to rank by what is worth reading. Whether that difference can hold up commercially, and what it does to public debate, is the subject of Suzuki’s session.

A Scientist Who Built a News Company

Suzuki did not come up through Silicon Valley. He holds a PhD in complex systems science, which studies how many interacting parts (cells, economies, ecosystems, populations) behave together in ways none of them shows alone.

Most aggregators are tuned by engineers watching click-through rates. Suzuki thinks about a news platform the way he was trained to think about an ecosystem, with feedback loops that can run away when each part chases its own incentive.

Before SmartNews he wrote “Namenaka no Sekai” (なめらかな世界). The book argues that many of the lines people draw, between self and other, online and offline, one community and the next, are leftovers from older information architecture, and that technology can make them more porous. He calls the result a “smooth society”.

The Filter Bubble Problem

Eli Pariser coined the term “filter bubble” in 2011. An algorithm tuned for engagement shows people what they already believe, because agreement holds attention longer than disagreement. Readers drift into separate versions of events.

Two fixes get tried. Editors can pick stories by hand for balance; that scales badly and brings its own bias. Or the recommender can be tuned to push opposing views; people tend to dig in when shown content they dislike, so this often makes things worse.

Suzuki’s diagnosis is different. Most algorithms model a user as a lone individual with fixed preferences. SmartNews models readers as participants in a system that keeps changing. Its machine learning scores each article on source credibility, factual density, topical diversity and other quality measures, and optimises for what the company calls “quality discovery” rather than engagement.

Studies of SmartNews’s US audience have found it more politically balanced than virtually any other major news aggregator, with a roughly even split of liberal and conservative readers.

Information Quality as Infrastructure

SmartNews’s mission statement is “Discover and Deliver Quality Information to the People of the World”. The company has put heavy investment into natural language processing and machine learning that judge a story’s originality, its sourcing and what it adds to public understanding, before the story reaches a reader.

The environment those systems work in is deteriorating. AI-generated articles are multiplying. Local newsrooms keep closing. State and non-state actors run campaigns through social media. In that setting, a news aggregator’s ranking decisions carry more weight than they did ten years ago.

SmartNews runs in the United States and in Japan. Japan’s large outlets still practise balanced reporting and public trust in institutional journalism is comparatively high, so the two markets give the company, and Suzuki, a live comparison of how the same product behaves under different media cultures.

Complex Systems Thinking in the Age of AI

Suzuki’s academic training reaches past news. As AI takes on more decisions in finance, healthcare, government and content creation, the question of how to keep those systems aligned with what people want gets harder.

A complex systems scientist does not ask “how do we make this algorithm fair?” on its own. The question is how the algorithm interacts with human behaviour, institutional incentives, markets and cultural norms. Fixes at that level usually mean changing the feedback loops, not adjusting a single parameter. Suzuki applies the same method to AI governance and content moderation.

Suzuki at Tech for Impact Summit

The 2026 theme is “Beyond Boundaries: Building 2050 Together”. Climate change, AI governance and inequality cross sectoral and national borders, and the programme is built for leaders who work across them. Suzuki’s session sits alongside the summit’s conversations on AI deployment, digital rights and the institutions needed to keep technology accountable.

Other speakers include former Minister Taro Kono on Japan’s digital transformation policy; Cardano founder Charles Hoskinson on decentralised infrastructure; GLOBIS founder Yoshito Hori, who gives the keynote on entrepreneurial leadership; Kathy Matsui, general partner at MPower Partners, on impact-driven venture capital; Monex Group’s Jesper Koll on financial markets; and Sota Watanabe of Startale and Ken Shibusawa of Commons Asset Management on Web3 infrastructure and multi-generational stewardship.

What Executives Take From It

Every organisation in the room, in technology, finance, media or government, both consumes and produces algorithmic information. Its decisions depend on what reaches its leaders through those channels.

Suzuki’s argument, backed by SmartNews’s own numbers, is that a company can rank for quality, hold a politically mixed audience in one of the most polarised media markets on earth, and still grow past 50 million downloads. Executives deploying AI in their own decision-making will hear what that took.

---

The Tech for Impact Summit 2026 takes place on April 26 in Tokyo. Seats are limited and allocated by invitation. Request your invitation to join Ken Suzuki and other global leaders in technology, media, and impact.