If we could slow down AI development, I would favor it.
I would like more time to understand what we are building and what it will mean for us. But I don’t believe China will stop because we do. A pause on our side would not necessarily be a pause for humanity. That complicates the question; it does not, by itself, establish that accelerating is safer.
I do not believe the Chinese regime is evil, or that an artificial superintelligence (ASI) developed in China would necessarily be bad. My preference for building it in America comes from the ideals I want to shape it, and how I believe those ideals can help us overcome our instincts.
I still lean toward accelerating, even though slowing down would be my preference. The reason is my fear about humanity’s course without AI, and I am not entirely comfortable with that judgment.
I find the absolute certainty on either side difficult to take seriously. People who are 100% accelerate or 100% slow down do not seem to me to be thinking rationally or carefully enough. There are enormous risks in moving forward. There are enormous risks in what happens while we stand still. I don’t see how anyone can weigh both and come away without doubts.
All-or-nothing thinking is deeply human. We are animals at heart. My view is that we evolved to see things myopically and to generalize because it helped us survive. Quick judgments and simple categories were useful. They still feel natural, even when the problems in front of us demand something much more careful.
I think this is probably the biggest problem with modern civilization. We have extraordinary power, but we still approach our disagreements with those same instincts. Our tendency to reduce complicated questions to opposing camps may be our destruction.
This is where my belief in American ideals comes in. The freedom and morality I value include a dedication to doing what works and asking what serves everyone, even when that means looking beyond the immediate interests of our own country. I believe those ideals can help us overcome the instinct to put our group first, whatever the consequences for everybody else.
The postwar international order is an example of what I mean. At Bretton Woods in 1944, the United States and its partners helped establish institutions based on the belief that economic cooperation could promote peace and prosperity. Alongside that economic project, the United States has played a major role in protecting international shipping; the Navy describes its mission as preserving prosperity and keeping the seas open and free.
America benefited from that system too, and maintaining it has been costly. What I admire is the willingness to sustain something whose benefits extend far beyond America. I see an understanding that other people’s prosperity can be part of our own, rather than something we have to defeat. That is the kind of thinking I want reflected in ASI.
I also worry that the America I believe in is dying: that we are losing our commitment to those ideals. In particular, I worry about the appeal of Marxism and socialism to younger Americans. I do not want those ideologies to provide the foundation for ASI; I want it grounded in the American ideals of freedom, practical judgment, and broadly shared prosperity that I have described. That is a disagreement about political ideas, not where people come from.[1]
That adds another urgency for me. I want ASI built in America while the values I admire can still shape its development. An American address alone would guarantee nothing; the culture doing the building matters.
But neither that preference nor my fear of losing those ideals settles the question of whether we should accelerate. I still have to make the case that doing so offers humanity a better chance of surviving. Wanting America to get there first is not enough.
My reason for leaning toward acceleration remains that I worry humanity will destroy itself unless we put AI in charge sooner rather than later. I hope intelligence beyond our own can help us get past the limitations we keep failing to overcome ourselves. The values we build into it matter because I want it to overcome our destructive instincts, rather than inherit them.
That is a hope, not a certainty. It does not make the risks of AI disappear. It is why I cannot treat slowing down as an obviously safe choice, however much I would prefer more time.
I believe this decade is the most important one in all human history. We need to think about it with more care than choosing a side and becoming certain we are right.