The question of work
Efficient — for whom?
The same word that sells a surveillance camera sells a self-checkout, a robotaxi, and a warehouse robot: efficiency. It's always worth asking the next question — efficient for whom?
Here's the honest picture, because honesty is the point. Automation is being deployed at real, accelerating scale. Some of the people who built the gig economy a decade ago are now watching it automate around them. And yet — so far — official employment in many of these jobs is still flat or growing. Both things are true. What's coming next is the part nobody can promise.
- 1,000,000+
- robots now in Amazon's warehouses, nearing its ~1.5M human workforce
- Amazon / CNBC
- ~500,000
- Waymo robotaxi rides a week — doubled in a year, with a target of 1 million by end of 2026
- Waymo / TechCrunch
- −13%
- the relative drop in employment for early-career workers (ages 22–25) in the most AI-exposed jobs since 2022, even as their older peers held steady — the first hard signal
- Stanford / ADP payroll data
- 54,836
- U.S. job cuts in 2025 that employers themselves attributed to AI — though some economists call AI a convenient cover for ordinary cost-cutting
- Challenger, Gray & Christmas
Sector by sector
What's happening — and the honest counterpoint
Driving → robotaxis
What's happening
Waymo's paid rides doubled in a year to ~500,000 a week across a growing list of cities, and it's aiming for a million. In the metros it serves, third-party earnings data (Gridwise) shows driver pay falling while it rose nationally: San Francisco −6.9%, Austin −5.3%, Los Angeles −4.7%, Phoenix −3.8%, against +1.0% nationwide.
But, honestly
Honest counterpoint: Gridwise itself cautions it's too early to pin this on robotaxis alone — pulled-back driver bonuses and seasonality matter too. The official U.S. outlook still projects driver jobs to grow this decade, and GM shut down its Cruise robotaxi unit entirely. The displacement looks real, but the cause isn't yet settled.
Amazon & the warehouse
What's happening
Amazon passed its millionth warehouse robot in 2025. A leaked internal plan, reported by the New York Times, described automating ~75% of operations and avoiding roughly 160,000 U.S. hires by 2027 — and 600,000+ by 2033.
But, honestly
Amazon denies the framing, calling the documents “incomplete and misleading,” and notes warehouse employment actually tripled over the automation decade. The dispute — leaked projections vs. corporate denial — is itself unsettled.
Delivery
What's happening
DoorDash, Uber Eats, and Instacart run on millions of gig couriers — and sidewalk-robot and drone fleets scaled up through 2025.
But, honestly
For now, every delivery robot and drone combined is a tiny fraction of delivery capacity, much of it small pilots; Amazon's Prime Air drones have struggled, and the official outlook still projects delivery-driver growth. The threat here is mostly about what's coming, not what's here.
Self-checkout & the cashier
What's happening
Cashier is one of America's biggest jobs (~3.3 million, ~$15/hour) — and the one facing the largest projected decline, which the government ties directly to self-checkout. Walmart says it cut labor hours per transaction 35%.
But, honestly
Even here it's nuanced: many shoppers prefer the speed, and some chains — Booths, Dollar General, Target, even Amazon's “Just Walk Out” — have rolled self-checkout back over theft, frustration, and the fact that it quietly turns the customer into the unpaid worker.
Offices, support & software
What's happening
The newest front is white-collar. Salesforce says its AI agents let it cut customer-support staff from about 9,000 to 5,000, and its CEO says AI now does 30–50% of the company's work. Across the economy, the hiring slowdown has hit entry-level office and coding roles first — the door that young workers walk through.
But, honestly
Honest counterpoint: it cuts both ways. Klarna publicly walked back its AI-first support after quality slipped and began re-hiring people; the U.S. government still projects software-developer jobs to grow ~18% this decade — AI as a tool that raises demand, not only replaces it. White-collar exposure is high; white-collar displacement, so far, is narrow.
What the data shows
The clearest early signal: driver pay
Most of the disruption is still a question about the future. But in the handful of cities where robotaxis already run, the first hard numbers are in — and they point the same direction.
Driver pay in the cities robotaxis serve
In the four metros Waymo serves most heavily, third-party data from Gridwise shows rideshare-driver hourly pay falling over the year to mid-2025 — while nationwide it rose. Gridwise cautions it's too early to attribute this to robotaxis alone — reduced driver bonuses and seasonality play a part too — but the gap is consistent across every robotaxi city. Each bar shows the size of the year-over-year change; only nationwide pay rose.
Source · Gridwise Analytics, 2025
The honest debate
Will it be a catastrophe — or just a churn?
The big forecasts disagree by a hundredfold, because they measure different things. Goldman Sachs estimated generative AI could expose ~300 million jobs worldwide — but said most would be complemented, not replaced. The World Economic Forum projects 92 million jobs lost but 170 million created — a net gain. MIT's Nobel laureate Daron Acemoglu thinks only ~5% of tasks will be profitably automated this decade.
History offers a famous warning against panic: when ATMs arrived, the number of bank tellers went up, not down, because cheaper branches meant more branches. A similar surprise is unfolding in trucking — despite a decade of “driverless trucks will end driving jobs” headlines, and real driverless freight now running in Texas, the number of U.S. truck drivers has actually risen, from about 1.7 million to over 2.2 million. The honest counter is that past automation hit physical work, while AI reaches into white-collar and cognitive jobs faster and wider — and the first hard signal is already here: early-career workers in the most exposed fields have seen their employment slip relative to everyone else, a sign the change may arrive first as jobs not created rather than jobs destroyed. Nobody knows for certain how far it runs. That uncertainty is exactly why the people affected deserve a seat at the table now — not an apology later.
One more honest twist: for routine, robotic automation, the heaviest burden has historically fallen on lower-wage, less-educated workers — disproportionately women and people of color. But generative AI may reverse that, exposing better-paid, better-educated office workers the most. The lesson isn't that one group is safe. It's that a transition this big needs to be managed for people, not just for efficiency.
The part the spreadsheets miss
There's a cost that doesn't show up in a productivity number: connection. The U.S. Surgeon General has warned that a lack of social connection raises the risk of early death by more than 60% — comparable to smoking 15 cigarettes a day. Research finds that even brief, “weak-tie” exchanges — with a cashier, a barista, a driver — measurably lift our sense of belonging. Those are precisely the human moments being engineered away.
The sociologist Allison Pugh calls it “connective labor” — the work of seeing another person and making them feel seen. A task can be automated. “What can't be automated,” she writes, “is the relationship between two human beings.” A city is made of those relationships. It's worth deciding, on purpose, how many of them we're willing to trade for convenience.
“Driverless taxis are flooding an already competitive market and taking money from human drivers. I do believe my income is being impacted.”
“I felt monitored and analysed as if I were a machine. It was a kind of cage.”
Close to home
What it means for Detroit
Detroit has lived this story longer than almost anywhere. The city that put the world on wheels also gave it the moving assembly line — the original bet that a machine could do what a person used to. A century later the same question is back in new forms, and it's landing on a workforce with less cushion to absorb the shock.
Detroit's median household income is about $39,600 — roughly half the national figure — and a large share of the city's workers earn their living in exactly the jobs now being automated: driving and delivery, cashiers, retail and warehouse work, customer support and office roles. When a national trend reaches a city like this, it isn't an abstraction. It's the rent.
Waymo is already here
Waymo began mapping and testing driverless cars on Detroit streets in late 2025 and says it's preparing a robotaxi service — the same technology now linked to falling driver pay in the cities where it already runs.
WXYZ; Axios Detroit ↗Drones over Pontiac
Amazon launched Prime Air drone delivery in Pontiac in late 2025 — one of only a handful of U.S. sites, and the first test of automated delivery in metro Detroit.
DroneXL ↗The warehouse bet
Amazon's fulfillment center on Detroit's former State Fairgrounds was pitched with about 1,100 jobs — the same kind of warehouse work the company's million-plus robots are designed to do with fewer people.
Bridge Michigan ↗The auto transition
Auto is the most robot-intensive industry on earth — about a third of all U.S. industrial-robot installations. GM cut roughly 1,000 jobs in late 2024, most at its Warren technical center, and Stellantis offered Detroit-area buyouts in 2025 as the costly, automation-heavy shift to electric vehicles reshapes the plants.
IFR World Robotics; Bridge Michigan ↗Detroit didn't just take it
Detroit also wrote the counter-story. In 2023 the UAW struck all three automakers and won the right to strike again if a plant is closed — and forced GM to bring its EV battery workers under the union's national contract. Automation is a question communities can still answer.
UAW; CBS Detroit ↗
Michigan jobs in the path of automation
How many Michigan workers hold four of the occupations most exposed to automation and AI, as of 2023. These are among the most common ways people in the state earn a living — and the ones most often named in automation forecasts. The point isn't a prediction; it's the scale of livelihoods the question touches, right here.
Source · U.S. BLS — Occupational Employment & Wage Statistics (May 2023)
None of this is anti-technology. It's a plea to choose on purpose — to make sure the tools we adopt serve the people who live here, instead of the other way around.
Show up when it counts
Decisions get made in rooms.
We'll tell you which room, and when.
The most powerful thing a resident can do is be present before the vote. Subscribe and we'll email you ahead of every City Council session, Police Commissioners meeting, and public hearing that shapes our city.