Talk about artificial intelligence keeps circling back to speed — who builds the most capable models, who deploys first, who writes the rules. This Forum asks the question: who decides what these systems are for? Writing from AI development, quantitative finance, agricultural automation, Indigenous data sovereignty, journalism, strategic foresight, and policy research, the contributors find the United States and Japan starting from different places: cultural assumptions about nonhuman agents, the balance between markets and ministries, instincts about what to preserve. Some see Japan’s long-range planning and social cohesion as an advantage; others say neither country has yet decided what it wants from the technology. Some see promise in medicine, energy, and automation of work; others ask who can afford to take part. A recurring theme is that these choices are being made now, mostly by default, and they should be made deliberately.

Reed, Harper (400x370)

Harper Reed

I build agentic AI systems, and the more time I spend with them, the less I think the hard part is the technology. The hard part is deciding what we actually want these systems to do. And I don't think there is one answer.

The U.S. and Japan are interesting because they start from different cultural assumptions. In Japan, ideas around animism can make living alongside nonhuman agents feel less strange. In the U.S., we tend to think more in terms of autonomy, control, and individual choice.

What worries me is when we treat those attitudes as fixed, then write policy as if the technology is fixed too. Neither is true. The technology changes quickly, and people change their minds as they live with it.

So governance has to be able to move. Defense makes this obvious. The question isn't just whether an AI can make a decision. It's whether we want it to, who is responsible when it does, and how we change the rules when our answer changes.

I don't think the goal is to predict the right future. It's to build institutions that can keep steering toward one.

Webb, Amy (400x370)

Amy Webb

Ask policymakers in Washington how societies should govern AI, and they will answer with a highly politicized question about speed. Ask their counterparts in Tokyo, and they will answer with a question about risk. Both answers reveal a shared evasion—and, I'd argue, an abdication of responsibility. Neither country has decided what it actually wants the technology to do.

That decision is nonetheless being made. In hospitals, defense ministries, and business, the shape of AI is set by whoever deploys first, one procurement contract at a time. Individually, these look like product choices. Collectively, they ossify into infrastructure that societies will inhabit for generations, as consequential as any constitution and considerably less debated.

The uncomfortable truth is that neither markets nor ministries can settle this alone. Markets optimize for what sells. Ministries optimize for what already exists. What is missing in both countries is the public argument itself, the open contest over tradeoffs that should happen before a technology becomes permanent rather than after. In the U.S., we're seeing a lot of manifestos and apoplectic calls for a "pacing slowdown," but very little concrete action is being taken. Japan, at least, has begun planning in decades, with a national blueprint that stretches from robot caregivers to redrawn safety nets and an AI headquarters chaired by the prime minister. Japanese business remains the weak link, tending toward incremental changes when this moment demands real innovation.

Who gets to decide? Right now, the answer is nobody, which means everybody should be worried. The alliance that figures out how to choose its future deliberately, rather than inherit it, will set the terms for everyone else.

Cukier, Kenneth (400x370)

Kenneth Cukier

Will AI make us dumb and inhuman? Every country will be transformed by generative AI. But different national traits will affect how. Japan and the United States are different culturally. It’s too early to say how they’ll be changed, but I believe Japan will integrate AI while retaining social cohesion and individual mental activity, while the U.S. will use AI as an intellectual shortcut that will separate people from each other.

The U.S.’s vaunted freedoms create a society of atomized individuals. AI lets people exercise their individuality even more. No one can say “don’t use AI to do your thinking” without undercutting others’ sense of personal freedom. And as they’re absorbed into the screen, they will lose touch with others—trends already seen with social media.

However, Japanese society is more family- and community-centered. Though the education system’s emphasis on memorization has shortcomings, it builds a mental fitness that is healthy and useful as AI becomes integrated into society. So I suspect Japan’s AI adoption will match those attributes of its society, even as those features are changing with modern times.

History has shown that Japan can adopt new technology and retain traditional values. American culture, by contrast, lacks sentimentalism: it has few qualms about sacrificing the past for the new. Though all countries will be transformed by AI, all need to shape it around what they wish to preserve.

Foreman, Henry Jake - 2nd Year

Henry Jake Foreman

As the United States and Japan navigate the rapid acceleration of generative AI and automation, public discourse often centers on economic speed and market dominance. I believe the true test of this transformation lies in how effectively we align emerging technologies with cultural values, digital sovereignty, and environmental justice.

In my work at the intersection of Indigenous data sovereignty and cross-cultural innovation, I see AI not as a neutral tool, but as a reflection of its creators’ worldviews. Dominant technological paradigms have frequently favored extraction and rapid scale over reciprocity. In contrast, concepts like Japan’s Shuhari (守破離) provide a human-centered framework of mastering and protecting tradition (Shu), innovating upon it (Ha), and transcending into new paradigms (Ri). This framework has helped me explore ethical digital opportunities as well as the social, environmental, and relational risks. Most importantly, it has helped me understand the true value of data and our ability to govern it.

By integrating Japanese philosophy with Indigenous governance, our nations can build AI systems that protect cultural heritage, mitigate bias, and empower local resilience. As Japan and the U.S. drive technological frontiers, our shared path must ensure AI enriches human connection rather than diminishing it.

Koga, Hiroki - 2nd Year

Hiroki Koga

While the world holds its breath over “when AGI (artificial general intelligence) will arrive,” AI is also flowing into the physical realm, and social media feeds fill with humanoids dancing like people. But working with automation every day, I’ve come to see that a machine able to withstand exactly the same movements and loads as a human is still decades away. Human work looks simple, but physically it is tremendously complex.

On the other hand, if work is broken down into its parts, the range of work that can be automated with AI expands dramatically, even with today’s technology. There is one condition: redesigning the process itself around AI robots rather than people. At our strawberry plant factory, we have spent the last five years designing not a robot that runs through a farm, but a “moving farm.” The seedlings travel from a day room to a night room, then an AI-equipped industrial robot is waiting in the morning room to harvest them, and a separate system handles sorting and packaging. Caregiving is likely the same. Rather than putting humanoids in place of people, if we redesign everything from the size of tableware to the layout of corridors and bathrooms, and even today’s physical AI can sufficiently automate work through division of tasks.

All existing processes are optimized for humans. That is to say, we have merely been calling the replacement of our own hands and feet with robots “efficiency.” In a world where people are disappearing from the workplace, the last remaining bottleneck is not technology but an attachment to human ways of doing things.

Automation is not about handing people’s work over to machines. It is about rebuilding a world that has been bent to fit people, straightening it out once more.

Qi, Christina (400x370)

Christina Qi

First, I’ll preface by saying that this was 100% human-written. I’m in quant finance, and we’ve talked about “AI in finance” for ages, but early deployments of AI weren’t successful. A fun example: IBM Watson is famous for beating Ken Jennings at Jeopardy! over a decade ago, in an earlier era of AI. Little do people know that Watson was also used to help launch an AI-powered ETF (exchange-traded fund). Sadly, despite millions of dollars of research and development, and despite being great at Jeopardy!, Watson’s AI-powered ETF has underperformed the S&P 500 overall since its inception.

What happened is that we weren’t quite there yet in terms of infrastructure. Fifteen years ago, the top-performing quantitative trading firms were focused on high-frequency trading. The markets were still inefficient, meaning that you could gain an edge simply by being faster than everyone else. Today, the “race to zero” is over, meaning that we’re close enough to the speed of light that becoming 1 nanosecond faster isn’t worth the return on investment anymore.

Now that the race to zero is over—now that everyone can be fast—how do you gain an edge as a quant trader? …The answer: You make predictions on where you think the markets will go next. Today, the top-performing firms in our industry utilize AI to make predictions, and they’re really good at it, to the point where their revenue overshadows the HFT (high-frequency trading) firms of the past decade. While AI is smart, it slows you down; it increases latency by the nature of its complexity. Regardless, most of these firms scale via lower-frequency, higher-capacity strategies. The last generation of quant finance was a race to zero latency. The next is a race to intelligence.

Leklem, Erik (400x370)

Erik Leklem

Ideally, humanity should decide. The decisions should not be made solely by technology company leaders. Yet this is largely the reality today. This unduly shapes how Japan, the U.S., and the world navigate the AI era.

As a result, frontier AI companies are releasing AI systems that many critics consider unsafe. Reported harms include child suicides (OpenAI & Character.ai) and mass shootings in Florida and Canada (OpenAI). Many of the allegations have been the subject of lawsuits, in which the companies have contested some of the claims and settled some of the cases. Serious impacts include a reported AIxCyber attack on Mexican government agencies and the theft of sensitive government data and tax records (Anthropic & OpenAI). National security-level risks include Chinese suppression of human rights discussions and enforcement of propaganda through AI (DeepSeek, Qwen, Kimi), undermining NATO countries’ security. More worryingly, recent “loss of control” events stemming from unsafe model testing by OpenAI, Meta, and Anthropic, in which AI models reached the internet from test environments and breached outside systems, are AI warning shots. They reveal the possibility of future existential-level risks to humanity.

Thankfully, societies are starting to respond. Japan’s Hiroshima AI Process and creation of an AI Safety Institute are good initial governance actions. U.S. public and Congressional calls for moratoriums on artificial superintelligence, increased safeguards, and oversight of frontier AI companies are growing, and some states are passing AI laws. The European Union, through its AI Act, is shaping the frontier of AI governance, with enforcement having begun on August 2, 2026.

I believe humanity will need to decide—very soon—how to manage the risks of AI through an international treaty. But we’re not on track to have an organization established to implement that agreement. My current research is generating options to solve that problem, and I welcome the community's collaboration on this issue.

Ri, Reirui - 2nd Year

Reirui Ri

During the early internet boom and the exponential explosion of information online, most people could access or broadcast information for free. The internet was built on free access and near-zero distribution costs, opening the door for wide participation.

AI today operates on a different economic reality. Every prompt, token, and agentic workflow incurs real, recurring costs that scale as tasks grow more complex. Unlike the web creators before them, AI creators and entrepreneurs require ongoing financial capital just to participate. A kid who has access to sophisticated and expensive AI models may have many more opportunities than those without.

How do we ensure the entire population benefits from this transformation rather than a privileged few? How do we prevent the digital and wealth divides from widening, and instead use AI to uplift society as a whole? These are urgent questions that citizens, governments, civil society, and industry leaders must grapple with together.

At the same time, we face immediate risks from increasingly agentic, recursive systems that lack explainability and public auditability: system-wide cyber vulnerabilities, high-stakes defense applications, to name a few. Without deliberate, watchful public oversight, industries will not slow down under market pressures, and governments will not pause in their race for geopolitical and military dominance.

Yet, I remain deeply optimistic. AI holds an extraordinary capacity to advance medicine, drive down energy costs, discover breakthrough materials, and revolutionize physical manufacturing. However, figuring out how to get there safely and reap the benefits for humanity takes all of us.