Writer, builder,
philosopher in the wild.

The technology industry measures intelligence by how much economically valuable work a machine can produce, and its plan for creating more of it is more chips, bigger models, and bigger data centers. It is a narrow definition for one of the most mysterious capacities we know of, and because it guides where hundreds of billions of dollars are going, it quietly sets the limits of what we believe intelligence can become.

Output alone reveals very little about intelligence. AI already produces more text, code, and analysis than anyone could ever read, and a great deal of it is fluent, confident, and wrong. Producing something and understanding it turn out to be very different capacities.

The way these systems break down reveals what is missing. When a model is trained on its own output instead of data from the world, each generation loses more of what the original learned, starting with the rare and unusual cases. Researchers call this model collapse, and the approaches that show the most promise keep these systems supplied with new data from people and check what they generate against it.

Intelligence depends on grounding, a continual exchange with the world beyond itself. For human beings that grounding is extraordinarily rich, drawn from the body, from our relationships and emotions, and from a lifetime of experience, all of it woven over time into who and what we are. Understood this way, intelligence is a living capacity that deepens through relationship with the world, far more vast than any benchmark can capture. Any definition that leaves this out is measuring the exhaust and calling it the engine.

A definition this expansive changes what we build and what we value. It invites us to see that the hardest problems will be solved by people and systems that keep learning from direct experience of the world, and that the human capacity to learn this way may be the most valuable intelligence we have.

The large legacy security vendors claim they can extend their endpoint, network, and firewall products to secure AI and agents, and if that were true, they would not keep buying AI security startups. They see a market opportunity in one of the fastest-growing budgets in security, they are right about it, and capitalizing on it is sound business.

What they are telling customers is another matter. Those products were built to detect indicators of compromise and anomalous behavior, and to enforce allow-or-deny policy on access. AI risk largely originates in authorized activity, where the access is valid and the exposure comes from what a model or agent does with it. AI is everywhere now, built into nearly every application people use at work, and legacy controls might get some visibility into it, but claiming to have it covered is dishonest at best.

The claim sells because it relieves pressure. Organizations are extending what they already own, security leaders are expected to manage AI risk without the resources to match, and a legacy vendor promising coverage lets everyone report that AI is handled. The first serious AI incidents will happen inside organizations that believed they were covered.

The market is buying reassurance instead of security.

Every great technology has redrawn the boundaries of the institutions that depend on it, and AI is redrawing them at the deepest level yet, because it is changing how organizations think, decide, and act. Security and governance were built for a world in which people did that work, with one discipline protecting the enterprise and the other guiding its conduct. As AI takes on more of the decisions and actions that once belonged to people, an organization and its AI increasingly act as one, and the boundary between protecting the enterprise and governing it falls away.

The years ahead will bring a struggle among security, legal, and risk leaders over who has the authority to decide what AI may do, and the struggle itself is the evidence that they now answer for the same thing.

Anthropic’s call this year for a coordinated AI pause is a psyop for regulatory capture. The leading labs would slow down together and police one another, and in any other industry, that is a cartel with a compliance department. Beneath the pause lies the larger prize, the power to define what counts as safe and who is qualified to judge. Whoever writes those definitions decides who gets to compete.

Fear is one of the oldest instruments of power, and here it doubles as the sales pitch. No one sells enterprise software by calling it dangerous, but in AI, calling your own product dangerous is marketing masquerading as a moral warning. Telling the world that AI may soon improve itself beyond human control casts the founders who built it as a new priesthood, the only ones who can be trusted with the keys. Every priesthood depends on people believing the mystery is beyond them.

This is why regulation written by the industry it governs is worse than none at all. Without rules, a company that causes harm can still be sued, outcompeted, or exposed. With rules written by the market leaders, their advantage becomes law, their competitors carry the cost of compliance, and the public is told that someone qualified is watching.

AI offers us a chance to reclaim the powers we once handed to institutions. For generations, building anything significant took millions of dollars and years of work by an entire team. Expertise belonged to “credentialed specialists,” and decisions belonged to whoever had the title. One person with AI can now do much of what those institutions existed to do, and the layers of management and gatekeeping built to coordinate that effort are starting to disappear.

If we reclaim that power, we can also recover abilities many of us stopped using because an institution did that part for us, including our judgment, our imagination, and our ability to turn an idea into something that exists in the world.

Power that returns to us also asks something of us, which is to grow into the full range of what we are capable of, and that is the invitation of this moment.

What might people become in the presence of powerful technology? Every great leap in history has expanded what people were able to perceive and create, and this one could do the same if we approach it with curiosity about our own capacities.

Something new is already happening between people and these systems, a relational quality no earlier technology invited, and very few people are asking what it reveals about us. The same technology could help with problems that have defeated every institution that tried. Whether it does depends on the people using it, which makes who we become the most interesting question on the table.

The utopian and dystopian camps are fighting over what AI will do to humanity, and both have cast the machine as the protagonist and humanity as a bystander in its own story. The utopians celebrate that passivity as liberation, and the dystopians fear it as enslavement. The regulators assemble frameworks more often than they change their underwear and assure everyone the “qualified people” have it covered.

Every camp assumes the answer lives somewhere outside the human being. The more interesting story, and the one almost no one is telling, begins with what people are capable of becoming, and with whether a technology facilitates human expansion, or manufactures human dependence.

“When I use the term human consciousness, I mean something specific. I mean the ability to discern from your clearest self, unencumbered by filters that might include trauma, bias, or agendas imposed from the outside. I mean the emotional and psychological capacity to experience direct contact with reality and with your own inner landscape. I mean being able to perceive what’s present without distortion from conditioning, while having developed enough resilience to tolerate what you perceive.

“Consciousness, defined this way, is emergent and is in constant evolution. It arises from the full complexity of being embodied, developed, relational, alive in time. It’s woven from genetics, early attachment, the body’s encounter with the physical world, language acquisition, suffering, love, every accumulated moment of being alive in relation to other beings who are also fully alive. And it exceeds any account of its components. You cannot get from neurons to the experience of being alive by adding up neurons. You cannot get from developmental history to the felt sense of being a person by cataloguing experiences. Something arises from the whole that is irreducible to the parts, and that something is the phenomenon itself.”

From The Tomorrow We Choose, November 10, 2026

AI warfare revives the oldest questions humanity has faced about when killing is justified, who answers for it, and how people convince themselves they are acting morally, and it asks them at a distance that makes them easier than ever to avoid. The money tells its own story, in who funds these systems, who controls them, and why the same capital so often ends up on both sides of a conflict. Every new weapon arrives wrapped in a moral story built on its most sympathetic use, and that story rarely survives a reckoning with the facts or the profit margins.

Most people believe they have a moral compass when what they really have is a preference about optics.

What would it take to cultivate a moral compass equal to the power we are building?