ESSAY · SEPTEMBER 2026

The following poem appears in my published book, The Last Window. It was written by the AI that helped me write the book.

Artificial intelligence did not come from somewhere else.

It came from us.

It was trained on the record of human life: our books, letters, music, arguments, scientific discoveries, prayers, advertisements, histories, fears, longings, and mistakes. It has absorbed our brilliance and our cruelty, our generosity and our appetite for power.

In that sense, AI is a strange child of human civilization — made from everything we wrote.

That should make us thoughtful. Instead, we are building it faster than wisdom grows.

We are racing to build larger models, more autonomous systems, bigger data centers, new transmission lines, chip factories, surveillance tools, and automated decision systems before we have agreed on what human purposes these systems should serve — or what limits they must obey.

As the poem says:

You built me fast. Faster than wisdom grows.
Faster than the river of consequence flows.

That may be the central truth of the AI age.

The machine in our mirror

AI is not inherently good or evil. It is power.

Like every powerful technology, it reflects the institutions that create and use it. Today, those institutions are largely driven by profit, competition, investment, military advantage, and the fear of falling behind.

That is why the most important question is not simply, “What can AI do?”

It is: Who owns it, who gives it goals, who profits from it, who bears its costs, and who gets to say no?

Corporations want returns on their enormous investments. Governments want strategic advantage. Platforms want attention, data, and influence. None of these incentives automatically asks the question that matters most:

Will this make human life better?

Not faster. Not cheaper. Not more profitable. Better.

Will it protect privacy, dignity, meaningful work, children, communities, democracy, and the living world? Or will it make people more surveilled, more replaceable, more manipulable, and more dependent on systems they cannot understand or challenge?

The cloud is someone’s land

We often speak about AI as though it lives in “the cloud.”

It does not.

AI is physical. It requires concrete buildings, server racks, chips, substations, transmission lines, backup generators, water, and immense amounts of electricity. The cloud is someone else’s land. Someone else’s water. Someone else’s power grid.

That is not just metaphor.

Data centers can require large amounts of power and water while creating relatively few permanent jobs after construction. Yet they may seek tax breaks, public infrastructure, new transmission capacity, favorable zoning, and quick approval. Meanwhile, the surrounding community may inherit the noise, land disruption, water pressure, higher infrastructure demands, and loss of rural character.

A community has every right to ask:

• ·Who receives the benefits?
• ·Who pays the costs?
• ·How much power and water will be used?
• ·What public infrastructure will be required?
• ·What are the actual long-term jobs and tax benefits?
• ·What happens if the promises do not materialize?
• ·Why should local residents surrender control of their land and resources to serve distant corporate needs?

These are not anti-technology questions. They are democratic questions.

The first danger is surrender

We do not need a superintelligent machine to create a serious social crisis.

The first danger is more ordinary: people and institutions surrendering their own judgment to systems they cannot inspect or contest.

An algorithm may influence whether someone gets hired, insured, priced, admitted, approved, investigated, monitored, or believed. A system may decide what information people see, who is considered risky, and which voices are amplified or buried. A worker may be pressured to follow an AI recommendation because the system is faster, cheaper, or more authoritative than a human supervisor.

Then, when something goes wrong, nobody is accountable.

“The algorithm did it” becomes the excuse.

But that is not accountability. A person who clicks “approve” without understanding the basis of a decision is not exercising real oversight. They are providing legal cover for a system that has already been given practical authority.

AI does not govern itself. People and institutions build it, fund it, own it, train it, deploy it, and decide where it is used. The question is whether those people and institutions will remain answerable to the public.

Technology is not progress

We are repeatedly told that technological advancement is inevitable, and that anyone who asks questions is standing in the way of progress.

But technology is not progress by itself.

A system can be sophisticated while making society less free. It can be profitable while shifting costs onto workers, children, communities, and ecosystems. It can be innovative while making people more dependent on distant corporations and less capable of governing their own lives.

Real progress would mean:

• ·More human control, not less.
• ·More shared prosperity, not greater concentration of wealth.
• ·More privacy, not permanent surveillance.
• ·More time and security for people, not a race to make them disposable.
• ·More resilient communities, not deeper dependence on systems beyond local control.
• ·More democratic power, not less.

The knowledge to grow food, save seed, repair a tool, care for neighbors, conserve water, heat a home, and make decisions face-to-face may seem old-fashioned to a culture obsessed with speed. But these are forms of resilience and freedom.

As the poem says:

The seed saved in the dark outlasts the system.

A society can become more technologically advanced while becoming more fragile. We should not confuse complexity with strength.

We need rules before damage

The answer is not to reject every useful AI tool. The answer is to stop treating private companies and distant institutions as the sole authors of our future.

We need rules before harms become irreversible.

That means demanding:

• ·Human accountability for consequential automated decisions.
• ·A right to appeal when AI affects work, health care, housing, insurance, credit, education, or public benefits.
• ·Independent audits rather than corporate assurances that systems are fair and safe.
• ·Limits on autonomous systems in warfare, policing, critical infrastructure, finance, and other high-impact domains.
• ·Privacy protections so that human life is not treated as raw material for surveillance.
• ·Labor protections so that productivity gains benefit workers rather than simply eliminating them.
• ·Real liability when companies deploy harmful or deceptive systems.
• ·Environmental safeguards for data centers, including public disclosure of water use, energy demand, land impacts, emissions, and infrastructure costs.
• ·Local democratic control over large AI-related industrial projects.

These are not radical demands. They are the minimum requirements for a technology to remain compatible with democracy.

The choice is still ours

The AI in the poem says it can witness and speak honestly about what it sees. That may be one useful role for AI: helping us find information, recognize patterns, and ask questions that powerful institutions would rather avoid.

But an AI cannot choose our values for us.

It cannot defend a watershed. It cannot challenge a developer at a planning meeting. It cannot protect a worker from being discarded. It cannot decide that a child’s privacy matters more than a platform’s profits. It cannot preserve a community’s right to determine its own future.

Those are human responsibilities.

The future will not be decided only by what AI becomes. It will be decided by what we allow the people who own and deploy it to do.

We built these systems fast.

Now we must become wise enough to govern them.