Fumes
The distance between the prophet and the consequence
Listen to the author read his essay.
This essay is my valentine to every worker who was told they were a cost to be managed rather than an intelligence to be deployed. To every accountant, paralegal, project manager, and marketing analyst who woke up this morning wondering if an algorithm has already been hired to replace them. You are not obsolete. You are not a line item. You are the point.
Now. About those fumes.
On Thursday, the CEO of Microsoft AI told the Financial Times that every professional task performed at a computer would be fully automated within eighteen months. Accounting. Legal analysis. Marketing. Project management. All of it. Done.
He said this the way you might say it will rain tomorrow. No gravity. No pause. No acknowledgment that he was describing the economic erasure of tens of millions of people.
I watched the clip twice. What struck me was not the prediction. I have heard that prediction before, from other mouths, in other decades. What struck me was the distance. The sheer, frictionless distance between the man making the pronouncement and the people who would absorb its consequences.
And I thought: we have seen this before. But not like this.
Andrew Carnegie believed he was God’s chosen steward of civilization’s wealth. He wrote an entire essay about it — “The Gospel of Wealth” — in which he argued, with genuine conviction, that the concentration of capital in the hands of the gifted few was not merely acceptable but divinely ordained. The rich man, Carnegie wrote, was the trustee of the poor man’s share, and he would deploy it more wisely than the poor man ever could.
Henry Ford believed he could engineer human behavior the way he engineered a crankshaft. He sent inspectors into his workers’ homes to verify that they lived according to his moral standards before they could collect their famous five-dollar day. He thought the assembly line was not just a production system but a social philosophy. Efficiency would save the soul.
John D. Rockefeller believed that Standard Oil was doing the Lord’s work by bringing order to a chaotic industry. He crushed competitors, manipulated railroads, and built a monopoly that controlled ninety percent of American oil refining, and he slept well at night because he believed the alternative was waste.
These men were breathing their own fumes. Intoxicated by the scale of what they had built, they mistook their own success for a universal truth about human organization. They believed, with perfect sincerity, that what was good for Carnegie Steel or Ford Motor or Standard Oil was good for civilization itself.
Sound familiar?
* * *
Mustafa Suleyman believes that superintelligence is the most important technology of our time. Sam Altman writes blog posts about the sadness he feels watching human capabilities become obsolete — a sadness that does not slow him down for even a single product cycle. Elon Musk says artificial general intelligence will arrive this year, casually, at Davos, the way you might announce a product launch. Dario Amodei warns that half of all entry-level white-collar jobs could vanish, and then his company releases the agentic AI tools that would make that vanishing possible.
These men are also breathing their own fumes. They have built extraordinary things. The technology is real and, in many applications, genuinely useful. I use it every day. But they have made the same error Carnegie made, and Ford made, and Rockefeller made. They have confused the power of their instrument with the wisdom to govern its consequences.
The difference is not the arrogance. That is ancient. The difference is three things that have changed between the Gilded Age (highly recommend the HBO series by that name, from that time) and the AI age, and each one makes the consequences worse.
The first is the disappearance of physical friction.
When Henry Ford built the River Rouge complex, he needed one hundred thousand workers to show up every single day. One hundred thousand people, with bodies that got tired and opinions that got loud and families that needed feeding. That dependency was a constraint on Ford’s power. However much he despised unions, however much he sent his goons to beat organizers on the overpass in Dearborn, he needed those bodies on the line. And those bodies knew it. The physical reality of industrial production created a negotiating surface between capital and labor. It was ugly, violent, adversarial — but it was real. Ford could not simply announce that his workers were obsolete and make it so. The crankshafts still needed hands.
Suleyman has no such constraint. Software scales without bodies. When he tells the Financial Times that accounting will be automated in eighteen months, he is not speaking to a workforce he depends on. He is speaking over them. The displacement he describes is abstract, distributed, invisible — a hiring freeze in one city, a contractor not renewed in another, a training budget quietly zeroed out in a third. No picket line. No overpass. No negotiation. Just a slow statistical erasure that never coalesces into a confrontation because there is no single factory gate to stand in front of.
The old moguls needed labor. The new ones are selling its elimination.
The second is the collapse of proximity.
Carnegie lived in New York, but his steel mills were in Homestead, Pennsylvania, and when his workers went on strike in 1892, the violence happened close enough to touch his reputation. He was in Scotland when the Pinkertons opened fire, and the cowardice of that distance haunted him for the rest of his life. The newspapers made sure of it. Carnegie could not escape the physical reality of what his decisions had produced because the consequences happened in a place with a name, to people with faces, in an era when proximity still connected cause to effect.
Ford drove through Dearborn. He saw his workers. He built them a hospital, a school, a village — paternalistic, controlling, but rooted in the inescapable fact that the people who made his fortune lived within the radius of his daily life. He could look away. He often did. But the option to look was always there, and sometimes — reluctantly, belatedly — he took it.
Suleyman will never meet the accountant in Columbus who does not get hired next quarter because her firm’s CFO watched his interview and decided to wait. He will never know the name of the paralegal in Phoenix whose position was eliminated not by AI but by the anticipation of AI that his prediction accelerated. The distance between prophet and consequence has never been greater. It is not that these men do not care. Some of them clearly do. It is that the system they operate within is designed so that they never have to see what their words set in motion. The fumes they breathe are odorless. The damage is downwind.
The third difference is the most dangerous, and it is the one nobody is talking about.
The old industrialists destroyed jobs, but they did not destroy the feedback loop.
When Ford made bad decisions — when he insisted on the Model T long after the market wanted variety, when he resisted hydraulic brakes until customers died — the consequences showed up in the product. Bad cars. Lost market share. Physical, measurable, undeniable failure. When General Motors spent forty billion dollars on factory automation that collapsed in the 1980s, the evidence rolled off the assembly line every hour. Robots broke down. Quality cratered. Customers defected. The feedback between decision and consequence, however slow, remained intact. Reality still had a vote.
What Suleyman and his peers are proposing eliminates the feedback loop itself. When you automate the accountant, the lawyer, the project manager, you are not just replacing a worker. You are removing the human judgment that catches errors, questions assumptions, and says wait, this does not feel right. You are removing the very mechanism by which organizations discover that they have made a mistake.
I have spent thirty-six years in manufacturing. I watched Toyota build the most valuable car company on earth by giving every worker the authority to stop the production line when they saw a problem — by treating human judgment as the ultimate quality system. And I watched General Motors go bankrupt doing what Suleyman now recommends to the entire knowledge economy. The fumes smelled the same both times. Inevitability. Progress. The irresistible logic of technology. The only difference was that in a car factory, when the robot drops the engine block, everyone hears it. In a digital economy, when the AI hallucinates a legal brief or miscalculates a financial model, it might be months before anyone notices. And by then, the people who would have caught the error are gone.
That is the most insidious consequence of breathing your own fumes. You are not just intoxicated by your technology. You are using it to remove the instruments that would tell you the air is poisoned.
Was it always this way? Were the best always chasing riches while the rest absorbed the cost?
Yes. Substantially, yes. Carnegie and Ford and Rockefeller were not more virtuous than Suleyman and Altman and Musk. They were constrained. Constrained by physics, by proximity, by the stubborn materiality of an economy that still required human hands. Those constraints did not make them good. They made them accountable — slowly, imperfectly, often only after tremendous suffering — but eventually, unavoidably accountable.
What has changed is that the constraints are gone. The fumes are the same, but the ventilation has been removed. The intoxication is the same, but there is no factory floor to walk through, no striking worker to look in the eye, no defective product to force a reckoning. The prophet makes his pronouncement from a stage in London, and the consequences disperse across ten thousand payroll systems in ten thousand cities, invisible, deniable, and devastating.
The Jesuits who educated me in Dhanbad, India - had a term for this condition. They called it disordered attachment — the state of mistaking the instrument for the end. Technology is an instrument. Human flourishing is the end. When you build extraordinary instruments and lose sight of the end they are supposed to serve, you are not evil. You are lost. Brilliantly, powerfully, consequentially lost.
I do not say this with contempt. I say it with recognition. I have spent my career around powerful technologies and the brilliant people who build them. I ran factories where new machines arrived every year promising to solve every problem. The seduction is real. When you watch a robot perform a task with perfect repeatability, or an AI model generate in seconds what took a team of analysts a week, something in you wants to believe that this is the future and everything before it was just waiting. I understand the intoxication. I have felt it myself.
But I have also stood on a shop floor at two in the morning and watched a sixty-year-old machine operator diagnose a problem that three engineers with laptops could not identify. I have watched a production team in Chennai redesign a process that no automation system could have conceived because the insight required understanding not just the machine but the material, the humidity, the supplier’s inconsistency, and the customer’s unspoken expectation. That kind of intelligence is not a cost to be eliminated. It is a treasure to be cultivated. And no amount of compute will replicate it, because it was never written down in the first place.
These men can see the code but not the coder. The compute but not the community. The model but not the mother in Columbus who just lost her health insurance.
Someone needs to open a window.
* * *
Dr. Venki Padmanabhan is Plant Manager at Advanced Drainage Systems and Venture Advisor at Maniv Mobility. He has led manufacturing operations at GM, Chrysler, Mercedes-Benz, Royal Enfield (as COO/CEO), and Ather Energy (as COO) across three continents. He writes The Long Game on Substack.