AI Won’t Take Your Job.

But your boss will use AI to justify what they were planning to do anyway.

Photo by Growtika on Unsplash

Alright, confession time.

I use AI every damn day. Last month, it chewed through a fifty-page contract for me in four minutes flat. I sat back, grinned like an idiot, and felt a rush that was probably the closest I’ll ever get to enlightenment. I’ve built companies in three countries. I know exactly how much it sucks to read a fifty-page contract line by line. Four minutes. I’m sold.

So when I say the AI panic is bullshit, I’m not some outsider waving a pitchfork at the robots. I use this stuff all the time, and I’ve watched — up close — what actually goes down when companies start bragging about their shiny new AI toys.

And here’s what I’ve noticed: the story they tell about AI and the story that’s actually happening are two completely different stories.

The public story is all disruption, transformation, and the unstoppable march of progress. You hear it at conferences with mood lighting and speakers who say ‘ecosystem’ with a straight face.

The real story? It’s way simpler. And honestly, it’s kind of ugly. Like, ‘don’t turn the lights on’ ugly.

AI isn’t coming for your job. Your job is getting axed because a bunch of suits in a boardroom decided it was good for their stock options, and AI is just the best excuse they’ve ever had. The robot didn’t fire you. The robot just made the bullshit press release easier to write.

Let me walk you through how we landed in this shitshow. Once you see the pattern, you can’t unsee it. And once it’s seared into your brain, maybe — just maybe — we can stop bullshitting ourselves and actually fix the damn thing.

Let Me Take You Back to a World That Sounds Made Up

Picture this. It is 1955. Dwight Eisenhower is the president. Elvis is about to change everything about music. And the United States government is taxing the top income bracket at 91%.

Ninety-one per cent.

Yeah, I know. Breathe. Maybe grab a drink.

Here’s the bit that should make every modern economist sweat: the economy was on fire. OECD countries were cranking out 4% growth in the 1950s, almost 5% in the 60s. Historians call it the Golden Age of Capitalism, and for once, they’re not even being sarcastic. Corporate tax rates? Over 50% for most of those decades. GDP still grew like crazy. Compare that to the limp 1.8% in the 2000s, when everyone was chanting the gospel of low taxes and trickle-down magic.

Now, let’s be real — because honesty is in short supply in these debates — the high taxes didn’t single-handedly cause the boom. Post-war America had a cheat code: Europe was rubble, US factories were the last ones standing, and people had been waiting years to buy anything that wasn’t rationed. The stars were aligned.

But here’s what those high taxes definitely did produce, whatever else was going on:

Corporations had nowhere to invest their money except in their own companies.

When the alternative to investing in your workforce is handing half your profits to the government, suddenly “let’s build a world-class research lab” starts looking very attractive. And that is exactly what they did.

They built Bell Labs, which gave the world the transistor, the laser, information theory, Unix, and cellular technology. They built Xerox PARC. IBM Research. They hired thousands of PhD scientists and basically said, “Go think.” Figure things out. We’ll keep the lights on.

They built real career ladders. A factory worker could actually become a middle manager, buy a house, send their kids to college, and retire with a pension. Not because companies suddenly grew a conscience. The system just made it the smart thing to do. If you can’t stuff your profits into your own share price, you put them into people and ideas instead.

That’s the trick. That’s the whole secret sauce of the so-called Golden Age. The system made investing in the future the path of least resistance.

And Then Someone Had a Very Clever Idea

Then the 1980s showed up with a shiny new theory. It looked great on a whiteboard and managed to screw things up for the next forty years.

The theory: corporations exist to maximise shareholder value. Everything else — employees, communities, long-term investment, research — is secondary to that single metric. Milton Friedman articulated it cleanly in 1970. Business schools fell in love with it. Regulators who should have pushed back didn’t. And slowly, the incentive structures of virtually every large company in the Western world were rebuilt around this single idea.

Taxes tanked. Stock buybacks got the green light. And here’s the real kicker: exec pay got chained to the share price like a junkie to his dealer.

So, if your bonus depends on the share price going up this quarter, what do you do with extra cash? Go on, take a wild guess.

You buy back your own shares. You shrink the pool, the earnings per share magically go up, the price jumps, and boom — your bonus lands. It’s not rocket science. It’s not even clever. It’s the most predictable scam in the world.

And it’s exactly what happened.

US companies spent an estimated $9.2 trillion buying back their own shares over the decade from 2012 to 2021. To give you a sense of the scale: that’s nearly 12 times more than was spent on buybacks in the entire period from 1982 to 1991. Since 2000, corporations have spent three times as much buying back shares and paying dividends as they did in the entire previous three decades.

$9.2 trillion. Not on R&D. Not on training anyone. Not on building the next Bell Labs. Just buying their own stock and — surprise, surprise — padding their own paychecks.

IBM is the case study that should be taught in every business school, precisely because it’s so painfully on the nose. From 1999 to 2019 — twenty years — IBM spent $176 billion buying back its own stock. One hundred and seventy-six billion dollars. Meanwhile, it missed cloud computing. It missed mobile. It missed the AI wave that its own researchers had pioneered decades earlier. The company that invented the future couldn’t see it, because it was too busy managing its share price.

This is not a technology failure. It is a governance failure. It is what happens when you take every incentive in a system and point it at a single number on a screen.

So Along Comes the Robot, Right on Cue

This is the part where the whole charade starts to feel downright insulting.

A 2026 Global Talent Trends survey asked 825 C-suite leaders whether they expected AI to lead to headcount reductions. Ninety-nine per cent said yes. Not 60%. Not 75%. Ninety-nine per cent.

That is not a technological assessment. That is a confession dressed up as a forecast.

Because here’s the thing about that same period: C-suites have been expanding, not contracting. The middle manager who spent fifteen years building institutional knowledge, knowing every supplier and every quirk of the system? Redundancy. Efficiency opportunity. The Chief AI Officer, the Chief Transformation Officer, the VP of People Experience and Culture Alignment, the Head of Strategic Synergy — these are investments. These are essential. These people have titles that require a full breath to say out loud.

Companies like Amazon, IBM, Meta, and Oracle have linked workforce reductions explicitly to AI in earnings calls, framing leaner workforces as “preparation” for the AI-driven future. The language is always the same: lean, efficient, agile, transformation. These words have been applied to layoffs in every technological transition since the advent of the printing press. What’s new is how good the alibi has become.

Klarna is my favourite example, because it has the rare quality of being both triumphant and immediately self-correcting. They laid off around 700 customer service workers, replaced them with an AI chatbot, and proudly announced the bot could handle two-thirds of conversations at a fraction of the cost. The tech press was ecstatic. Then, about a year later, customer satisfaction had dropped sharply. The CEO — to his credit, he was honest about it — admitted they’d “focused too much on efficiency and cost”, and that quality had suffered. Klarna began rehiring human agents.

Seven hundred people lost their jobs. The company spent a year figuring out what anyone who has ever had a frustrating chatbot experience could have told them for free. And then they started hiring humans back.

Meanwhile, the CEO was busy giving interview after interview about the glorious AI future, as if nothing had gone sideways.

The workers, meanwhile, still had rent to pay and zero patience for corporate fairy tales.

The Numbers Behind the Narrative

I want to be precise here, because the full picture is more complicated — and actually more hopeful — than the panic merchants on either side would have you believe.

The World Economic Forum projects that by 2030, 92 million jobs could be displaced by AI, offset by 170 million new ones created. Net positive of 78 million jobs.

But — and this is a very large but — that math only works if retraining programmes exist at scale. And right now, in virtually every country, they don’t.

Over 150,000 people were impacted by AI-driven layoffs in 2026 alone. More than 100,000 in 2025. At least nine companies — Accenture, Amazon, Citigroup, Dell, HSBC, Intel, Microsoft, TCS, UPS — announced AI-related layoffs affecting 10,000 or more employees each. These are not marginal adjustments. These are structural shifts.

And look at who is bearing the weight of it. The number of CEOs planning to cut junior roles specifically jumped from 17% in 2025 to 43% in 2026. Entry-level workers. People at the very beginning. People who haven’t had twenty years to build savings, networks, or institutional credibility to weather a restructuring.

In the 1950s, a young person starting a first job could reasonably expect their employer to invest in them. To train them. To give them a path. Not out of generosity, but out of rational self-interest, because the system made that the smart thing to do.

Today, that same young person is told to ‘upskill.’ To be a ‘lifelong learner.’ To stay ‘agile.’ The company, of course, won’t pay for any of this. They’re too busy shoveling cash into share buybacks. But hey, the LinkedIn posts are super supportive.

What This Looks Like on the Ground — In India, In Singapore, In Places Close to Home

Here’s where I want to bring this closer, because this isn’t just a story about American CEOs and American tax policy. I live in Goa. I’ve spent years building businesses in Singapore and the UAE. And the same story is playing out here, just with different costumes.

In India, the gig economy has grown from 7.7 million workers in 2020–21 to an estimated 12 million today, and it’s projected to reach 23.5 million by 2029–30. That sounds like growth. It is growth. It is also, if you look closely, one of the most aggressive reclassifications of labour in modern economic history.

We took people who might once have had jobs — with contracts, with insurance, with some kind of stability — and reclassified them as “entrepreneurs.” Entrepreneurs who happen to have no pricing power, no benefits, no collective bargaining rights, and no recourse when the algorithm decides to cut their per-delivery rate. Zomato and Swiggy delivery workers. Ola and Uber drivers. Dunzo riders were working in 45-degree heat in northern India, while the rest of us ordered cold drinks to our air-conditioned homes.

Oxford University’s Fairwork project assesses gig platforms on basic standards — fair pay, fair conditions, fair contracts. In 2024, they evaluated eleven major Indian platforms. Ola, Uber, and Porter scored zero. Zero. Not low. Not barely passing. Zero, on a ten-point scale designed to measure the most basic labour standards. No platform could guarantee workers a living wage. None would recognise a collective body or a trade union.

And what do we call this? We slap a shiny label on it: ‘democratisation of work.’ ‘Flexible employment.’ We wrap it in the language of freedom and agency because it sounds way better than what it actually is: dumping all the risk onto the people least able to handle it, while the platforms skim their cut and send happy updates to their investors.

Singapore tells the same story with a different accent. Singapore is the country that built its post-independence miracle on education, institutional quality, and the social contract of a genuinely developmental state. The CPF. Housing programmes that gave ordinary workers a stake in the country’s growth. It was, in its way, a version of the 1950s American model — a system designed so that the gains from growth had somewhere to go besides the top.

And yet. By 2024, Singapore had approximately 70,000 regular platform workers — almost entirely in ride-hailing and food delivery, concentrated in Grab and its competitors. A DBS study found that full-time gig workers were spending 112% of their income just to sustain themselves, with emergency savings covering less than two months of expenses. The median gross monthly income for a full-time Grab driver is between S$1,500 and S$2,500. Unchanged from the previous year, even as the cost of living increased.

Then there is the story that doesn’t make the GDP spreadsheet, but absolutely should. Singapore’s veteran diplomat Tommy Koh — not a firebrand, not a union organiser, a decorated establishment figure — stood up in 2019 and warned the government directly: we should not abandon displaced workers, he said, because “we don’t want more and more Singaporeans to become Grab drivers or, worse, to join the ranks of the angry voters.”

That sentence lands harder when you know who is becoming a Grab driver. Former bank executives. Retrenched PMETs — professionals, managers, executives, technicians — in their 40s and 50s, after twenty or thirty years in financial services or technology, finding that the jobs for which they are qualified have been restructured away. Their skills are real. Their experience is real. The market for those skills has simply been reorganised, and the reorganisation didn’t include them.

A former senior vice president of a bank, driving a Grab, is not a flexible entrepreneur exercising agency in the gig economy. He is a seriously underemployed person in a system that has failed to build a landing pad for the people it displaces. Singapore, to its credit, recognised some of this and passed the Platform Workers Act in 2024, which came into force in January 2025, requiring CPF contributions and work injury compensation for platform workers. It’s meaningful. It is, at minimum, an acknowledgement that “you’re an entrepreneur now, good luck” is not a labour policy.

But here’s the thing about that law: the workers who welcomed the CPF contributions are the ones thinking about retirement. The workers who are worried are the ones who need cash today. One driver told a reporter: “The delivery rider wants immediate cash.” That’s not ingratitude. That’s a person for whom the structural incentives of the system have produced a situation where saving for the future is a luxury, because surviving the present is already the full challenge.

This is the gig economy from the inside: not freedom, not entrepreneurship, just a daily cage match with an algorithm you didn’t build, on rules you didn’t write, with no union and no HR to call when shit hits the fan. The platform has investors. You have a phone and a delivery bag. Good luck.

What’s Actually Going On (The Honest Version)

AI is not the villain. I want to say this again, because I mean it.

AI is a tool — genuinely one of the most remarkable tools in human history — being deployed within a system specifically designed over forty years to extract value from workers and deliver it to shareholders. The tool didn’t design the system. The tool is just making the system more efficient at doing what it was already doing.

The gig economy did this before AI arrived. It took the employment relationship — with its obligations, its reciprocity, its social contract — and dissolved it into a transaction. AI is simply the next, more sophisticated phase of the same project. The entry-level analyst has been replaced by a language model. The junior lawyer was replaced by a contract review tool. The customer service team was replaced by a chatbot, then quietly rehired when the chatbot turned out to be terrible, in a story that repeated itself so often that Klarna eventually just admitted it out loud.

That system was not an accident. It was not the natural order of things, discovered by economists like gravity. It was built — through specific decisions about tax policy, regulatory frameworks, antitrust enforcement, and corporate governance — in the 1980s. Some of those decisions had genuine intellectual justification. Some of them were self-serving, dressed up in the language of economics.

The key thing I want you to take away from all of this is one sentence:

The problem isn’t that AI is powerful. The problem is that nothing in our current system creates an incentive to share what it produces.

That’s it. That’s the whole diagnosis.

In the 1950s, the incentive to share came partly from tax structure, partly from union density (which peaked at around 35% of American workers in the mid-1950s), and partly from a social contract forged in a Depression and a World War that nobody had yet had the comfort of forgetting.

None of those conditions exists today. So the productivity gains from AI are flowing to the people who own the machines. And the workers who are displaced — whether they’re mid-career bankers in Singapore, delivery riders in Mumbai riding through 45-degree heat, or entry-level analysts in New York who graduated straight into a restructured industry — are told, very encouragingly, to be more agile.

But Here’s Where I Get Genuinely Optimistic

I know how this sounds so far. I know you might be reading this on a Monday morning, already anxious, and thinking: great, another article telling me everything is broken. Thanks, very helpful.

So let me tell you the thing that I actually believe, underneath all the sarcasm and swearing:

The situation is fixable. Not because the technology will save us, but because the system was built by people, and people can rebuild it differently.

The mechanisms that once made corporations invest in their workers didn’t vanish. They were repealed. They were policy choices. And policy choices can be revised.

Here’s what that could actually look like.

Make reinvestment the attractive option again. The lesson of the 1950s isn’t “91% tax rates are magic.” The loopholes were real, the effective rates were lower than advertised, and the world is too global now for a simple replication. But the structural insight holds: when handing money back to shareholders becomes genuinely less attractive than investing it in workers and research, companies will invest in workers and research. A meaningful increase in buyback taxes — the current 1% is genuinely a rounding error — is the obvious first step. Several economists have proposed raising it to 4–8%. That doesn’t end buybacks. It tilts the incentive.

Share the gains from automation. Some European jurisdictions are already experimenting with this. When a company achieves significant productivity gains through automation, a portion goes into retraining funds or transition support for affected workers. This is not novel. This is the kind of negotiation that happened after every major technological shift in history — eventually, after enough disruption and enough political pressure. The question is whether we wait for the disruption to become catastrophic before we start negotiating, or whether we get ahead of it.

Extend real protections to gig workers, everywhere. Singapore’s Platform Workers Act is a start. Rajasthan’s Gig Workers Act is a start. But a welfare fund fed by 1–2% of platform transactions, while a meaningful first step, should be the floor, not the ceiling. The gig economy has spent a decade arguing that classifying workers as employees would destroy the model. The UK reclassified Uber drivers as workers. Uber survived. The model adapted. The argument that basic protections are existential threats to platform economics is, at this point, unsupported by the data.

Fund the retraining infrastructure that everyone agrees is needed. The WEF’s optimistic job numbers assume massive, scale-up retraining. Someone has to build that. The interstate highway system wasn’t built by Uber. The internet wasn’t invented by a startup in a garage. The foundational public investments of the 20th century — including the research environment that made Bell Labs possible — were public. The 170 million new jobs that AI could theoretically create require people who are prepared to do them. Preparation requires investment. Investment requires someone to make it.

Tell a better story. This is the one that people in policy circles always underestimate. The 1980s consensus lasted as long as it did because it came with a compelling narrative: freedom, markets, individual responsibility, wealth creation. It was simple. It was optimistic in its own way. It gave people something to believe in.

The alternative needs an equally compelling story — not one of fear and redistribution, but one of shared prosperity and long-term strength. Because here’s the historical truth: America’s most innovative decades were not its most unequal ones. The companies that invented the transistor, put humans on the moon, and created the foundations of the modern internet were operating in an era of higher taxes, stronger unions, and more aggressive antitrust enforcement. Those conditions did not prevent ambition. They shaped it. They gave it somewhere useful to go.

The Part Where I Ask You to Sit With This

I’m not asking you to believe that 1955 was perfect. It wasn’t. The Golden Age was golden for some people and not others, and the social contract of that era had exclusions written into it that we should never romanticise.

I’m asking you to notice something simpler: the choices that created our current system were choices. Not gravity. Not fate. Not the natural order of market economies, discovered by brilliant economists who had no personal stake in the outcome.

Choices. Made by people. And people can unfuck them.

AI is not going to take your job in the dramatic, sudden way the panic merchants suggest. What’s actually happening is slower, quieter, and in some ways more insidious: a steady restructuring of who gets to benefit from increases in productivity, using AI as the most sophisticated excuse the corporate world has ever had.

That restructuring has a human face. It’s the Zomato rider in Chennai who gets deactivated from the app for a rating he doesn’t understand, with no appeals process and no union to call. It’s the fifty-five-year-old bank VP in Singapore, with a mortgage and two kids in school, opening a Grab driver account because the job that was supposed to last until retirement was restructured away six months ago. It’s the twenty-three-year-old in any city who graduated into a market that has decided her entire entry-level job category is now an “AI efficiency opportunity.”

These are not inevitable. They are consequences of specific choices made in specific rooms by specific people within specific incentive structures.

The technology will do what we tell the system to reward.

Right now, the system rewards extraction. We could build one that rewards investment instead. The evidence from the decades when we did — imperfectly, with the wrong exclusions, with real loopholes, in a world we can’t simply recreate — suggests that what you get when companies are compelled to invest in people and ideas is something genuinely extraordinary.

Bell Labs gave us the transistor on a research budget funded partly by the structural incentives of a high-tax era.

Imagine what we might build if we stopped spending $9.2 trillion on share buybacks and started spending it on what comes next.

I think about that a lot. Especially when I’m four minutes into a contract review that used to take all afternoon.

The tool is remarkable. Let’s build a system worthy of it.

Gladwyn Lewis is a serial entrepreneur and three-time founder based in Goa, India. He has built and scaled teams across the UAE, Singapore, and India, and writes about political economy, institutional design, and the gap between what systems promise and what they actually deliver.