Banks are planning deep AI job cuts while Walmart promises AI will help — same tool, opposite stories

Information Technology 2 min 4 sources

The same week brought two opposite messages about AI and work: banks quietly preparing mass layoffs, and Walmart telling staff AI will improve their jobs. The technology is identical. What differs is who gets to keep what it produces.

Key takeaways

The same week, banks prepared mass AI layoffs while Walmart told staff AI will improve their jobs — identical technology, opposite promises.
A tool that doubles output can mean half the staff or twice the work; which one happens is decided by whoever captures the gain, not by the tool itself.
The growing anti-AI backlash isn't only fear of the unknown — much of it is people correctly sensing that the gains are real and aren't being shared with them.

Two stories ran past each other this week, and read together they say more than either does alone.

The same tool, two messages

Banks are laying the groundwork for mass workforce cuts as AI takes hold [1]. Walmart, meanwhile, is telling its workers the opposite: AI will improve your jobs, not steal them [2]. Same technology, two completely different promises about what it means for the people doing the work.

The detail that captures the moment: Andre Bonnick, a student at Warwick University, now rehearses for job interviews that aren’t with a person at all — he’s coached to use the right keywords and make “eye contact” for an AI-powered screening system that decides whether a human ever sees his application [1]. He’s already living inside the question the headlines are arguing about.

Who decides which story comes true

Here’s what the “will AI help or replace us” debate keeps missing: the tool doesn’t decide. A system that makes a team twice as productive can become half the staff doing the same work or the same staff doing twice the work — and which one happens is a choice made by whoever captures the gain. The bank and the retailer aren’t using different AI. They’re making different decisions about who keeps what the AI produces.

When the gain concentrates, the backlash arrives

That decision is starting to show up as politics. One economics columnist this week argued that anxiety about AI is set to generate a real political backlash — a coming “anti-AI populism” [3]. Its violent edge is already here: a 20-year-old in Texas was arrested for allegedly trying to burn down OpenAI’s headquarters and Sam Altman’s house, an anti-AI manifesto beside him [4]. It’s easy to file that under fear of the unknown. It’s more honest to notice that much of the anger is people reading the situation correctly — sensing that the productivity gain is real, and that it isn’t being shared with them.

The real question

So the useful question isn’t the one everyone’s asking. “Will AI help you or replace you” sounds like a question about technology. It’s actually a question about power: who gets to keep what the new tool produces? Answer that, and you can largely predict which story — Walmart’s or the banks’ — a given workplace is about to tell.

02 · Lesson · why it matters

A new tool never decides who wins from it — power does

The same machine can mean a better job or no job, a cheaper product or a fatter margin. Which one you get isn't set by the technology — it's set by who has the power to keep what it produces.

In one week, banks prepared to cut staff because of AI, and Walmart told its staff AI would make their work better. They are not using different machines. They are making different choices about who gets the gain — and that choice, not the silicon, is what decides whether the technology shows up in someone’s life as a raise or a layoff.

The tool is neutral; the split is not

Picture a tool that doubles what a team can produce. That extra output is real, and it has to go somewhere. It can become fewer workers doing the same work (the gain goes to the owners). It can become the same workers doing twice the work and getting paid more (the gain is shared). It can become a cheaper product (the gain goes to customers). The technology is identical in all three. What changes is who has the leverage to claim the surplus.

This is the oldest pattern under every “the machines are coming” story. The power loom, the spreadsheet, the shipping container, the ATM — each made someone vastly more productive, and in each case the fight that mattered wasn’t can the machine do it but who keeps what the machine makes. The answer was never read off the technology. It was decided by who held the power when the gain landed.

Backlash is the gain trying to rebalance

Which is why the anger is worth reading carefully rather than dismissing. When a productivity gain is real but flows almost entirely to the people who own the tool, the people whose work it replaced are not being irrational when they get angry — they’re sensing, correctly, that something valuable was created and they got none of it. “Anti-AI populism,” and even its ugly violent edge, is partly fear of the unknown and partly an accurate read of an unfair split. Treating all of it as ignorance is its own kind of blindness.

The half worth holding

It’s tempting to land somewhere clean and superior — the optimists are naive, the tech is just going to take the jobs — or its opposite, relax, every past panic was wrong, it’ll all work out. Both skip the actual variable. Whether it works out isn’t fixed; it’s being decided, case by case, by who has the power in each workplace and each country, and those decisions could go many ways.

And you are not watching this from the outside. You’re inside it — as a worker whose gain might be shared or pocketed, as a job-seeker already talking to an AI screener, as a customer, a shareholder, a voter who helps set the rules of the split. The humble move isn’t to predict the machine’s verdict, because the machine doesn’t have one. It’s to keep asking, every time someone says a new tool will help or hurt us: help or hurt whom — and who decided? The technology will not answer that. People will, and you’re one of them.

03 · Lab · your turn

Who Keeps the Gain?

Decide where an AI productivity windfall goes — profit, workers, or customers — and watch four different worlds appear from the identical tool, feeling that the outcome was a choice about power, not technology.