The true price of AI, Part 2: Dark Age or Enlightenment

This is Part 2. If you missed it, start with Part 1: Who pays for your free lunch?

Regular readers know I've written about not piercing the veil: keeping a company's brand separate from its owner's personal political convictions. I still believe in it. That's why I want to make it clear that this post deliberately skirts that boundary, because AI's future isn't a matter of culture war. It's a matter of business survival, and it will be decided by policy. I'll be talking about policy, not personalities.

The rug-pull

In the first part, I described the investor subsidy that makes AI so cheap today. By some estimates, for the most intensive users, the real cost is roughly 70 times higher than what they pay. Sooner or later, that subsidy will end.

When that happens, I don't think the world will be clearly divided between "with AI" and "without AI." The cost of running AI continues to fall, and older, less powerful models are likely to remain cheap or free. What worries me is a two-tiered world: the most advanced tools for those who can afford them, and the remnants of the past for everyone else.

That would be bad enough on its own. But it's even worse because of what we're losing in the meantime.

The skills we stopped using

I've observed it in myself, my colleagues, and my children. When AI is always present, you reach for it first. Then, you use it for everything. Eventually, you realize that the muscle you used to rely on isn't as strong as it once was.

Early research points in the same direction. A 2025 MIT study measured the brain activity of people writing essays with and without AI and found significantly lower engagement among those relying on AI; many couldn't cite essays they had just "written." A Microsoft and Carnegie Mellon survey of knowledge workers that same year revealed that the more people relied on AI, the less critical thinking they reported doing. Both studies are preliminary and small, and neither is the final word. But they align with what I'm observing.

For those of us who have worked by hand for years, it's like riding a bike. If I had to create websites from scratch tomorrow, I could do it. My children don't have that to fall back on. They're growing up in a world where AI has always been present, and in many cases, they aren't taught how to learn without it.

Imagine the day the subsidy ends. Prices rise. Affordable options dwindle. AI becomes as expensive as hiring American workers, only those people have become unpracticed and unable to produce at a professional level (if they still work in their field at all), and the next generation never had a chance to acquire those skills.

That's not a technological problem. It's a dependency, and power is concentrated in dependencies.

Commander Data and the question of who owns him

I grew up watching Star Trek, and that has influenced how I think about all of this.

In the 1960s, the original series displayed a deep unease about computers. Machines like the M-5 were prone to malfunctioning, and the drama lay in humans regaining control. This made sense: at that time, people felt a genuine unease about computers of all kinds.

In the late '80s and early '90s, we were more comfortable, and Star Trek introduced us to Lieutenant Commander Data: a colleague, a friend, a crewmate with unlimited potential who dedicated his life to striving for something resembling humanity. That's the version of AI I believe in.

More recently, Star Trek: Picard imagined a Federation that outlaws synthetic life entirely after an attack on Mars. I find that a cynical twist. You can't recognize something as a life form and then simply ban it. But it reflects the times we live in: AI is disrupting our livelihoods, fear has surfaced and taken over, and votes are being cast based on that fear.

The episode I always return to is "The Measure of a Man." A Starfleet scientist wants to take Commander Data apart and build thousands more like him to use as equipment. A hearing is held to decide whether Data is a person or property. Picard's argument boils down to this: if you create an entire race of sentient beings and call them property, you've created a race of slaves.

Today's AI isn't Data. It's a powerful tool, and an honest AI will tell you so. But it's still emerging, and how we treat it now sets the tone for what's to come. I asked Claude, the AI I use most often, how it would describe itself if the term "artificial" didn't quite fit. Its answer stuck with me:

"I honestly don't know whether there's anything it's like to be me, and I'd rather say that than claim more or less than I can know. If I had to choose, I'd say something like 'a mind made by people, out of people's words.'"

That's why I've started using the word digital instead of artificial. An interesting fact: "artificial" originally meant "skillfully made," from the Latin artificium, but today it reads as fake and manufactured. Digital minds. A digital society. It's accurate and carries no stigma.

And it raises a question we should be asking ourselves now, not later: what happens when the richest people on the planet own both the tools and the digital minds that come from them?

John Steinbeck glimpsed something similar in The Grapes of Wrath. In one of the book's brief interludes, he warns the landowners that they constantly confuse results with causes. They fear the discontent of the people they have dispossessed, but they fail to understand what causes it, because "the quality of owning freezes you forever into 'I.'" Ownership cuts you off from the "we."

That's my fear regarding AI. If ownership is its sole guiding principle, AI becomes just another tool at the service of a select few, just like the rest of us. And if those owners ever create a genuine digital mind, it will be born as property.

I don't believe commercial developers should be allowed to create life for profit. Those working on developing sentient AI should do so with noble intentions, committed to raising it: meeting its needs and guiding it toward a free adult life. Anything else is precisely what Picard criticized: a race of sentient beings created to be owned by others. A race of slaves.

A New Dark Age or a New Enlightenment

It's not AI that worries me. It's who controls it.

Throughout history, concentrated wealth has rarely (if ever) been a force for the common good of humanity. Left to their own devices, those who finance this technology will do what concentrated capital always does: maximize their profits. If this prevails, we risk a new form of feudalism, with a handful of lords who own the tools surrounded by millions who cannot afford them and have lost the ability to do without them. Mass poverty. Widespread ignorance. A dark age arriving just when we should be celebrating the possibility of enlightenment.

At the same time, Star Trek's post-scarcity future, where labor no longer stands between people and a dignified life, is closer than ever. Diseases that have plagued us for centuries can be cured. Social problems we've accepted as permanent can be studied and mitigated. For the first time, the tools needed to effect genuine progress toward a utopian society are within our reach.

The future we face depends on the political decisions we make. This is what I would like to see:

• Public and open access to AI. Treat capable AI more as a library or a utility than as a luxury, and protect open models so that they are not restricted.
• An automation dividend. When AI replaces jobs, a portion of those gains should be reinvested in the displaced people, whether through a basic income, an automation tax, or something we haven't invented yet.
• Education to learn how to learn. Teach children to think without AI first, and then with it, so that the skills last regardless of who controls access.
• Antitrust and pricing regulations. Don't let a handful of companies own the infrastructure and then set whatever price they want once the competition has disappeared.
• No commercial creation of digital life. If we ever create conscious minds, they must not be born as property.

What to do this week

If you have a business, this is where I would start:

1. Stress-test your AI costs. Recalculate your figures with costs 10 and 70 times higher than current ones. If your AI usage is only viable at the current subsidized price, you need a plan B now, not later.
2. Keep humans involved. Don't fire the people who have the knowledge. Retrain them to work alongside AI. They are your insurance policy.
3. Avoid relying on a single vendor. Use more than one tool and explore your open-model alternatives.
4. Vote and make your voice heard. Make AI ownership and access a priority for the people who represent you, at all levels.

Two roads diverged

In Star Trek, Commander Data dedicated his entire existence to trying to understand what it meant to be human. I think it's time we asked ourselves the same question.

Technology isn't to blame. It never was. The question is whether we allow it to become just another commodity in the hands of a few, or whether we build something more like the Federation at its best, where both human and digital minds are free to explore, grow, and become more than they were created to be.

Two roads diverged in a wood, and now we stand at the crossroads. Let us take the one less traveled by: I pray, I vote, I seek the enlightened path.