Below are Norm CEO and Founder John Nay’s opening remarks at the 2026 Central Park AI Forum.
…
There is no stopping. Resources will be deployed toward technological progress, one way or another. Thousands of years of evidence of this.
Whether it’s China, or a federated open source crew, or some other means: AI capabilities will advance. We’ll get further behind on solutions presuming they won’t.
AI infra is the biggest bet the U.S. has ever taken on one thing as a percentage of GDP (4% a year over the next 5 years). If we turn back now, we will crash the entire economy, and cede the win to other countries that are not democratically governed. Democratically-focused AI is the north star. I’ll come back to defining this.
It’s not easy to adapt to change but it's part of what gives life meaning.
Understandably, there’s a longing for stasis. I see this in myself all the time. I joke with my 4-year-old son and 2-year-old daughter that I want to freeze them. Just as they are today. Forever cute and precious.
Alas, change and growth marches forward.
In the context of AI change, people are lost.
Ezra Klein, who led the abundance agenda, wrote in the NYT a couple days ago that we should have AI stop writing code. That’s mainly what AI does now. Many Democrats want to dramatically inhibit AI, the only thing that can actually extend benefits to everyone.
I’m going to discuss a few different rounding errors both parties should internalize.
Redistribution of today’s stock of wealth is a rounding error compared to what we could generate by taxing future AI growth. And even setting aside redistribution and taxes, we can create net new value through fully automated free services to citizens. Not redistribution, but accomplishing outcomes.
Republicans are dismissing the need to proactively deal with how advanced AI could infringe on the liberties of individuals. Investing in safety is a rounding error compared to the costs of misuse of AI by bad actors. Investing in social safety nets are a rounding error compared to the social unrest that could ensue from unprecedented disparities.
It’s hard for anyone to grok these things because of the non-linear increase in AI capabilities. Asked in Sept 2025, the best Superforecasters in the world put the chance of AI solving a Millennium Problem (advanced math problems) at 1.7%. They underestimated almost everything else too.
Now that these things are coming to pass, people are in AI psychosis. This is understandable for the average person. Inexcusable for leaders.
Leaders have an obligation to stay measured in the fog of war and to stand for their actual ideals regardless of polling. Today’s “leaders” reverse engineer their policy positions from polling. Real leaders put forward such a solid and inspiring vision that it influences the polls. It shapes the narrative to move people toward a Pareto improvement to society.
AI is not popular? OK, death and disease are not popular either. What is popular? Curing all diseases, free services for all: education, health, financial, and legal services for free for all at no cost to others? That should be popular.
Let’s pause AI for sure. But not now. It's done basically nothing for us yet and we are at the early phases of its build out.
Let’s pause after it cures most diseases. Let’s pause after we figure out how to plug into our bureaucratic systems to unlock services provision to everyone. Let’s pause after we get some abundance. After we have a positive sum outcome that expands the pie.
And, potentially most importantly for the long-term, after we instill in AI the ideals of democracy.
Individual autonomy is core to American law, and maintaining that principle needs to become core to advanced AI. The capacity of a person to direct their own life, make independent choices, and act without unjustified interference is foundational to civil liberties and constitutional rights. We need AI to be centered on amplifying this for the transition to an increasingly AI-infused world to go well.
This is a goal all reasonable sides can rally around.
More powerful AI can give each human greater leverage to direct their life if we protect collective self-government (i.e., democracy) and individual self-government. People should shape law through democratic institutions, while enjoying fundamental rights that constrain what others can impose on them.
Miraculously, individual autonomy in America coexists with equal protection for the autonomy of others. Bridging this balancing act into a post-AGI world is the most important project of our time.
AI could curtail this in the hands of China, or even in our own hands. Alternatively, AI could expand the practical substance of individual autonomy.
Let’s focus on how to make the latter happen.
With frontier AI capabilities, each person could soon have access to financial, legal, government, and other expertise, creative tools, and productive capabilities not even currently available to the largest institutions. Each person could better understand their options, acquire skills, pursue discoveries, and challenge unfair decisions. AI could help people translate intention into action, which makes a wider range of lives genuinely possible, and therefore increases autonomy and meaning.
We need to lean into this in an increasingly AI-infused world by protecting access to AI, privacy, freedom from manipulation, and meaningful opportunities to contest decisions or change AI providers. People need to be able to delegate activities to AI while retaining the ability to redirect or revoke delegation. American law should continue to prevent one person’s or institution’s AI from undermining another person’s freedom.
Bringing AI agents into the fold of human law is critical to ensuring that AI is governed by democratically endorsed values, the most legitimate comprehension of societal values available. Other data sources suggested for AI alignment – surveys, humans labeling “ethical” situations, or (most commonly) the beliefs of the AI developers – lack an authoritative source of synthesized preference aggregation. Law is fundamentally grounded in verifiable resolutions via courts.
Humans are becoming increasingly dependent on AI. The question then is how to bake a dependence on humans into AI. We should use American law and its hundreds of years of application as the practical source of inspiration. Central command and control (in the AI, industry, or government sense) will not work in the messy plurality of reality.
AI far beyond human intelligence likely has no inherent economic or physical need for humans. Given this possibility, symbiosis between humanity and superintelligent AI should be deliberately developed. The most direct way to do this that can garner actual support across society today is to focus on instilling the individual human autonomy that’s core to American law as a primary goal of advanced AI. Increasing AI capability could expand each person’s capacity to choose and act, which would make extraordinary intelligence not just compatible with citizens’ self-government but a boon to it.
AI should be directed toward expanding what people can autonomously accomplish. In pursuing this, a test of AI progress should be whether people gain more control over their own lives as AI becomes more capable. That universally desired goal requires significant technical work and institutional norm-setting.
Today’s discussions will touch on all of these themes. And we have real leaders here today that will share their visions for how to make AI go well for everyone.
Think about the change between last year and this year's CPAIF. Now imagine the same non-linear change before next year. There’s a lot to figure out!