The Fork in the Road

Twenty-six essays ago this argument began at a factory in Fremont, California, where American workers GM had written off as lazy and ungovernable matched Japanese quality within months once Toyota treated them as intelligent partners in production. Same people. Different system. Opposite result.

Everything since has been evidence for one claim: the problem was never that American workers lack capability. The problem is that we built a machine that suppresses their intelligence and extracts the value they create in spite of it.

Thirteen essays established the suppression. Thirteen more established the extraction. Friedman supplied the moral permission. Buybacks supplied the mechanism. CEO pay supplied the incentive. Private equity industrialised it. The gig economy automated it. And the communities that produced the workers absorbed what was shed.

I want to end by telling you what I have actually seen, because the evidence in this series came from institutions and the conviction came from a floor.

I have watched a team leader named Ramon Hernandez take a chronic defect onto himself during a launch, gather data shift after shift, and trace it to a tool on his own team’s station — catching a fault before it ever reached a customer. Nobody instructed him. He had been formed to look.

I have watched a million dollars of inspection equipment installed to catch what nobody had been trained to do right, and switched off six months later because the operators had finally climbed the curve on their own.

I have stood at the top of a subcontracting chain on fifty acres in Tamil Nadu and been unable, thirteen years later, to name one person who built the plant.

And I have hired two men out of a homeless shelter, watched them flourish, formed them, and lost them both — because the wage I could offer was worth about two dollars an hour less than the thing across town.

The capability is real, it is already present, and every instrument we have is pointed the wrong way. That’s the argument.

Now the fork.

The AI moment

Artificial intelligence is the most powerful capability technology since electrification, and it is arriving at the exact moment when every extraction mechanism this series documented is fully operational.

That produces two futures. They use the same technology. They produce opposite outcomes.

Path one: extraction, accelerated

Deploy AI inside the current model and it completes the project Taylor started in 1911.

AI as surveillance. Every gig worker monitored in real time, every warehouse movement tracked by the second, every deviation flagged or terminated. The algorithm of Essay 10, supercharged — and note what it does not have. My tester could tell me a circuit had failed. It could not tell me the operator had never been taught to seat the connector. An extraction algorithm has no input channel for what the worker knows, because it was not built to have one.

AI as replacement. Every frontline role that can be automated, eliminated — not because it produces better outcomes but because it produces cheaper ones in the quarter the decision is made.

AI as wage suppressor. Real-time adjustment at a precision today’s platforms only approximate, capturing productivity gains that flow entirely to shareholders.

AI as buyback fuel. Automation savings redirected to repurchases. The Boeing pattern, applied everywhere at once.

This path is not speculative. It is the default trajectory of every incentive currently in place.

Path two: intelligence, amplified

The same technology does something else entirely if the premise changes.

AI as multiplier. An operator with twenty years of tacit knowledge works alongside a system that makes what he knows visible, shareable, and scalable. His judgment becomes more valuable, not less.

AI as the channel that was missing. The suggestion gap — Toyota’s hundreds of thousands of annual suggestions against the American trickle — is a channel problem. Build systems that solicit, process, and implement what workers see, and every operator has a route into the design of his own work.

AI as formation accelerator. The $774 per learner of Essay 7 becomes personalised development that adapts to what a person can already do. Not a compliance video. Not a screen with clickable parts. A supplement to a master, never a substitute for one.

AI as value-sharing. If gains from human-machine collaboration are shared with the people who produced them, AI closes the productivity-wage gap instead of widening it.

What decides which

The technology does not choose. The people who control capital allocation choose.

If executive pay stays tied to stock price, AI gets deployed for extraction. If governance stays captured by quarterly primacy, it replaces workers rather than developing them. If the policy environment keeps enabling unchecked buybacks, misclassification, and tax avoidance, AI’s gains flow exactly where every other gain in this series went.

But the evidence runs the other way, and it is not sentimental. Toyota outperforms GM. Costco outperforms Walmart. Lincoln Electric has not laid off an American worker for lack of work since 1948 and remains first in its industry. The firms that invest in worker capability beat the firms that extract from it, on the metrics the extractors themselves claim to care about.

The system, named plainly

There is a name for what the second path requires, and this is the right place to say it in ordinary words.

A lotus roots in the mud, rises through the water, and opens above the surface to the sun. Three media, in sequence, in the life of