Episode Details

If you prefer, watch the author express this essay to you.

———

Thanks for reading The Long Game by Dr. Venki Padmanabhan! Subscribe for free to receive new posts and support my work.

Stanford just proved what manufacturing has known for forty years.

In a paper called “Canaries in the Coal Mine,” researchers Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen used payroll data from ADP — the largest payroll processor in America, covering twenty-five million workers — to track what’s actually happening to jobs since generative AI went mainstream.

What they found should terrify every parent of a twenty-two-year-old.

Since late 2022, employment for software developers aged twenty-two to twenty-five has declined by twenty percent. Customer service workers in the same age group: down eleven percent. These aren’t projections. These aren’t forecasts. This is what already happened. Measured in actual paychecks.

Meanwhile, workers over thirty in the same occupations? Their employment continued to grow.

The canary is dying. But only the young canary.

Why? Because young workers in knowledge jobs carry what economists call codified knowledge — the kind of intelligence that can be written down, systematized, and therefore automated. Older workers carry tacit knowledge — the unwritten, experience-based judgment that AI cannot easily replicate.

When you’ve spent ten years debugging production systems, you don’t just know the code. You know which error messages actually matter and which ones you can ignore. You know that the system crashes every third Tuesday because of a batch job that nobody documented. You know which team member writes clean code and which one’s work needs a second look.

That’s tacit knowledge. It’s the knowledge you can’t Google. And it’s exactly what AI cannot replicate — because it was never written down in the first place.

Now. Here is the finding that changes everything. The one the headlines are missing.

The Stanford team didn’t just measure job losses. They measured the difference between two kinds of AI deployment: automation and augmentation.

In occupations where AI is primarily used to automate tasks — to replace what humans do — employment for young workers declined. That’s the twenty percent drop in software developers.

But in occupations where AI is primarily used to augment human work — to amplify what humans do, to make their intelligence more effective — employment actually grew.

Let me say that again because it matters more than anything else I’ll write this year.

Where AI replaces, jobs disappear. Where AI amplifies, jobs grow.

That is the entire argument of my book in one sentence.

———

I have been making this argument using manufacturing data for three years now. Let me connect what Stanford found in knowledge work to what we already know from the factory floor.

At Royal Enfield in India, I watched what happens when you deploy worker intelligence instead of replacing it. We didn’t eliminate workers. We built their capability — systematically, through training, through respect, through what I call the Three Opercula: Sanctuary, Ascension, and Crucible. We gave workers the conditions to think, the pathways to grow, and the challenges to prove themselves.

Profits grew twentyfold. Not because we bought better machines. Because we deployed better humans.

At General Motors’ Lansing Grand River Assembly, I watched what happens when you take a hostile, suppressed workforce and create conditions for intelligence to flow. Workers like Wally Vinton — belligerent, confrontational, ready to quit — who turned into the most innovative problem-solvers on the floor once management demonstrated that support was real. That plant won JD Power’s Gold Plant Quality Award. Not because of technology. Because of deployed human intelligence.

At Honda, I watched what