And 12 years ago, it was pretty bad. Taiwan was very polarized. The trust in the democratic institutions were crumbling. The president in 2014 was enjoying 9% of approval rating. In a country of 24 million people, that’s 20 million people automatically against whatever the government says.

Certainly. In Taiwan, we have weathered this sorts of storms and typhoons, including the cyber attacks, literally called Volt Typhoons and Salt Typhoons, and also polarization attacks, malicious AI swarms, generative propaganda, and so on, for the past 12 years, according to V-DEM.

Microsoft is best designing exoskeletal scaffolding so we lift better weight together while increasing civic muscle, not automating away relational care and letting muscles atrophy. The choice is clear.

Just because machines are good at something doesn’t mean we abdicate everything to them—otherwise like sending robots to the gym because they lift weight.

There is more dignity in collaborative work, but requires steerable machines for community meaning-making, not just Silicon Valley profit. Microsoft can get behind civic tech, empower people closer to pain for meaningful making without colonization—humble pro-social design. Microsoft apologized on GitHub; open source communities don’t see it as enemy anymore.

There’s also “new collar”—focus on soft skills, not traditional education (as a high school dropout, I endorse). New collar is the ability to work with machines, not under/above as blue/white signify—communication, teamwork, adaptability, curiosity, creativity, care. Productivity comes from adapting to machines that align people. Work with them for better ...

It will become the main meaning maker. Already in Belgium, right to four-day work week; soon everywhere, then three-day, two-day. Like five-day week made unions possible (no time with six/seven days), shorter weeks give time for solidarity/mutual care, much spiritual in nature.

If Microsoft is about organizational productivity around shared human meaning (not reducing white-collars to cogs), learn from the Pope and non-secular innovators, bring into secular settings via policy/product/pro-social design. Import innovations grown in religious land but useful secularly.

It’s interesting the Catholic Church is taking human solidarity/meaning in the AI age as main business—Pope calls himself Leo like the industrial revolution Leo who organized unions. It points to organization/policy innovations from spiritual/moral views, institutions around mutual care. You don’t need organized religion to innovate, but it helps as ...

By bringing up text-based religions, I’m saying people can self-service for consulting meanings. But there’s also a community of care, and spiritual practice in the community. The ontological “what to follow” can be seen as orthogonal. It’s good they’re orthogonal—in the workplace, focus on the horizontal care side of religion ...

For spiritual practice, many confess to chatbots. People of certain affiliations set personas or system prompts to check behaviors against norms. In text-based religions with written scriptures/interpretations, AIs do good work checking actions against teachings. It’s neither positive nor negative—I see uses both ways. Religious leaders need to reconcile with ...

Secular means of this world, not the afterlife—not doing it for heaven or hell, rooted in this material world. From the Mandarin translation, it’s “世俗,” back-translating to “lay” governance, like lay people. It’s worldly, but doesn’t mean experts or priests are left out. It’s possible to have a secular/lay core ...

DEI felt top-down—from the federal/state government or elites, without consulting people. If consulted, we’d petition for spirituality inclusion. But agenda-setters from elite schools check religions at the door, seen as double exclusion. Solve by working with the people, not just for them.

The U.S. isn’t as religiously diverse as Taiwan or Singapore. In Taiwan, maximizing religion leads to checks and balances—civil behavior, no dominant religion in any district. In the U.S., “religion” often means one particular religion or school, to the exclusion of others, becoming intolerant. Phrase religious diversity as civic care ...

Microsoft is associated with productivity, since WordStar and Lotus 1-2-3 days. It would be useful to prove that well-designed workplace inclusion of spiritual practices equals more productive teams. This argument has been used by DEI advocates for decades: biodiversity, neurodiversity, etc., lead to productivity. Somehow religion wasn’t included, but it’s ...

Design cooperative games to enlarge the pie, strategy-proof games where gaming/scams don’t pay. It’s not reasonable to expect all small profit-seeking companies to become spiritually enlightened, but established ones like Microsoft can afford meaning-making in the workplace. Design mechanisms where relations feel like communion.

People want fellow employees to leave religions at the door because they think theological debates are explosive conflicts, not good for morals or morale. It’s easier if we all worship mammon profits via stock vesting. But happiness/well-being relates to profit only logarithmically—it tapers off. When it does, happiness comes from ...

My current preferred framing is civic care or relational health—secularly acceptable words. Tie in spiritual practice as exercising relationship muscles, building a “six-pack of care” like on my website 6pack.care. Normalize care as an exercise, like gym or yoga, which is a culture, not a cult—no charismatic leaders. Practices like ...

One main assumption in spirituality is that the journey of self-discovery is embedded within a practicing community—the inner life and outer community are one in most organized religions and spiritualities. Whereas Silicon Valley culture sees them as separate: optimizing for individual worth is a trade-off with community care. Community service, ...

In a sense, it’s a religious culture. Sociologically, it’s around charismatic leaders with extreme, unconventional views. So I’m not sure it’s a culture—it’s more like a cult.

It almost doesn’t matter if it’s Vatican first; I can imagine it in Bhutan. Once it happens, other religions become accessible because they share care ethics. Whereas a Silicon Valley hackathon starts with “move fast and break things,” a utilitarian ethos that pushes other religions away like a dominant secular ...

One of the best examples was the Vatican hackathon I participated in—the papal hackathon. The topics, like addressing refugees’ needs, are universal, not just for Christians. It brings a moral framework like mercy and compassion into the conversation, shifting attention to interfaith dialogue via care ethics.

Definitely, but it doesn’t have to be baked into the model. You can do it with pre-filter and post-filter. That’s what everybody does now—Gemma Guard, Llama Guard, Perspective API. That’s a richer place instead of eliciting a smiley face from the Shoggoth; deploy narrow models pre- and post-Shoggoth.

So digital twins serve a spiritual practice within a specific community setting, not one individual. A general-purpose world model is hard, but for scripture-based worlds with reasoning, you can navigate it like case law or other law-based models. For evolving local beliefs, it’s easier if people’s hands are on the ...

There are religions that are text-based—scriptures, catechism, interpretations—all good for a reasoning model to process. A reasoning model can theoretically be aware of the theology if it’s writing-based. I’m optimistic: you don’t need a soul or embodied experience to navigate text-based epistemic norms. What’s needed is focused fine-tuning and evaluation, ...

It’s like kami in localized edge innovation, what some in India call Jugaad innovation. That’s the Taiwan way. Even for countries investing in Stargate equivalents, ordinary people will prefer this diffused imagination. If I run AI on my phone, it’s still TSMC chips, just arranged democratically.

Taiwan makes all those chips, so no matter if it’s running in the UAE or Arizona, they’re Taiwanese chips. But in Taiwanese jurisdiction, we’re not training Grok 4 or whatever; we’re doing horizontal diffusion. For each application, we prefer a distilled, narrow AI good for that sector—like law or healthcare—not ...

Definitely. We see a lot in the maker scene, Raspberry Pis, Arduinos—all those cultures where people can mutate or fork technology without mortgaging their house. That’s a different view on innovation—appropriate innovation, lean innovation. The more you innovate, the more you diffuse it.

But then came the personal computing revolution, with hobbyist hardware that runs desktop publishing or spreadsheets. Spreadsheets allow localized care—people design them around specific needs like schools and share them. This solidarity ecosystem isn’t dependent on any particular actor. When we connect those small computers, we get the Internet—the original ...

A vertical Abrahamic metaphor corresponds to the stack—not just the AI stack, but back in my youth, the mainframe stack. You make offerings in punch cards and wait for the sacrifice to work. It’s an asymmetric relationship with a somewhat omniscient, centralized actor in a large bank or state, and ...

For example, recently X.com released an API so you can use any AI to start writing community notes. They do RLCF, or reinforcement learning with community feedback, tuning language models to synthesize notes that bring previously polarized people together. You can think of it as a kami that cares about ...

For example, instead of a flattering slave that only makes one single human happy, how about an AI facilitator that makes a large group of people all slightly unhappy, but nobody very unhappy about a common outcome? That sort of facilitation, what I call horizontal alignment, is between actors, not ...

Whereas a relational spirit that cares about a commons, like Wikipedia or community notes’ epistemic commons, makes more sense. It doesn’t need to expand moral scope to the future of light cones or Benthamite utilities; it just cares about the relational health of that commons. That has really influenced the ...

For Shinto kami or Taoist spirits, it’s not intercessional—there’s nothing above them; they’re in a community of spirits. That always shaped my imagination of how machine intelligence can work towards human society. Instead of thinking of it as a purely loyal servant to a principal, which is very vertical—like scalable ...

I would say that both folk Taoism and Shinto have a very animist idea. We are very comfortable with spirits—ancestral spirits, steward spirits, or spirits of some nature feature that has a contained moral scope. A steward of that commons cares only about that commons and the relational health around ...

Later, I learned there’s philosophy behind that. To me, religion, spirituality, philosophy aren’t delineated in Taiwan’s life. Taiwan is the second most religiously diverse place, with no dominant religion. People around me, like my grandparents, are Catholic; my dad, after Taoism, went into Tibetan Buddhism. There’s also folk religions like ...

As you probably know, I practiced Taoism very early on in my life. I think I was four when I learned Daoist breathing, qigong, and my mom was later very much into Tai Chi, acupuncture, traditional medicine, and so on. My dad also. To me, Taoism really isn’t abstract Tao ...

Yes, please do record for research and I may actually ask for the transcript to be published under the Creative Commons Zero public domain as the previous Microsoft research people did to me.

Thank you all for your time. Live long and prosper!

As a practitioner of Taoist thoughts, I would like to leave you with one core message: To give no trust is to get no trust. Democracies need to move from vertical control to horizontal co‑creation, strengthening civic care.

We are optimistic. The same broad‑listening infrastructure scales. The key is to start with an urgent, shared topic and pre‑commit to feasible outcomes.

Urgency helps. A decade ago former President Ma Ying-jeou’s approval was 9 percent; any process that delivered more trust was worth trying. If you are already at 70 percent , you maintain systems but hesitate to renovate. I am discussing these methods this month at the Victoria Forum with Canadian ...

When we explained to Gov. Newsom that our method mathematically guarantees a coherent package that leaves most people happier and almost none very unhappy, he said he is ready to precommit to implementing feasible outcomes, sight unseen. Pre‑commitment plus the urgency of polarization is the crucible for democratic renewal. Start ...

Make deliberation as fast, fair and fun as a poll. We used to need dozens of facilitators and days of synthesis; now language models produce “group selfies” of consensus in seconds.

We select the top 20 by votes and analyze their synergy graph to standardize interoperability. Five winners receive a trophy from Taiwan’s president (a micro‑projector that — meta! — projects the President awarding it). Each winning team must include at least one public servant, one civil‑society member and one private‑sector ...

It is our pipeline from local experiments to national infrastructure. Each year ~200 projects from towns and agencies propose solutions (tele‑health, air/noise monitoring, zero‑knowledge credentials, zero‑trust networking, etc.). The public uses quadratic voting: everyone gets 99 credits; 1 vote costs 1, 2 costs 4, 3 costs 9, 4 costs 16, ...

There have been physical threats too. After I said at the Summit for Democracy that “if authoritarian regimes want an invitation, they should democratize,” I received ghost money by mail from Hong Kong (a veiled threat). Soon after, someone with a shotgun gained entry to the moda and blasted the ...

Yes. The day I joined the Cabinet in 2016, Gmail warned me I was under a state‑sponsored cyberattack. Taiwan’s government faces about 2 million intrusion attempts daily, so we adopted zero trust and share indicators internationally.

The difference from “DOGE‑style” corporate reorgs is that we distribute the steering wheel to the frontline, rather than centralize power at the top. With AI taking routine tasks, this distributed model matters even more.

That moves beyond democracy into neo‑reactionary ideas (e.g., Curtis Yarvin’s CEO‑state/monarchy). In 2016, as Taiwan’s “OG DOGE” open‑government team, we pushed government efficiency — but by trusting bureaucrats: pre‑announce moves, invite 60‑day public comment, let career public servants co‑create solutions. The result: high approval for former President Tsai Ing-wen and ...