Gig Workers Sue Uber Over Pay Algorithms as AI Cuts Freelance Job Postings 30%

The gig economy is getting hit from two directions at once. Delivery riders say the algorithms that assign jobs and set pay have quietly cut their earnings for years. Freelancers on platforms like Upwork and Fiverr are watching AI eat the exact tasks that built the gig economy in the first place.

The Algorithm Fight

In Edinburgh, food delivery riders working for Deliveroo, Uber Eats and Just Eat say their pay has fallen even though they're logging the same hours, according to The Guardian. A rider identified only as David, who's been delivering for seven years, told the paper he's making half what he made four years ago for the same workload.

Xabier Villares, an eight-year rider and lead organizer at the Workers' Observatory, a charity founded by gig workers with academics at St Andrews and Edinburgh universities, said the shift accelerated in the last three years. He used to work four or five days a week and cover rent. Not anymore.

The complaint isn't just about pay levels. It's about the black box. Riders don't know how the algorithm decides what they're offered or why the rate changes minute to minute, a system platforms call "dynamic pricing." The Workers' Observatory, which just secured a decade of research funding, has been running experiments to reverse-engineer how the pricing actually works.

This isn't just rider complaints with no backup. Research from the University of Oxford and Columbia Business School found that Uber's 2023 dynamic pricing rollout led to drivers earning "substantially less" per hour, according to The Guardian's reporting. That's a documented finding, not a grievance.

What's still unresolved is the legal claim. Drivers in the UK, the Netherlands and other countries this month filed a class action against Uber alleging they live in "constant fear" of a "soulless" algorithm controlling their pay and job allocation. That's an allegation in a lawsuit. It hasn't been proven in court, and Uber hasn't been quoted responding to the specific claims in these reports.

Asking a company to explain how its own pricing algorithm works isn't a radical demand. People who supply their labor to a platform have a reasonable interest in knowing the rules of the game, the same way anyone signing a contract wants to see the terms. Trade unions pushing to ban dynamic pricing outright are a bigger ask, and that's worth separating from the transparency demand.

The AI Fight

On the freelance side, the threat isn't opaque pricing. It's disappearing demand. A joint university study of nearly two million freelance job postings found demand for freelance writing dropped roughly 30% within eight months of ChatGPT's launch, with software development and graphic design not far behind, according to Forbes contributor Hammad Ali Nasir.

But the picture isn't pure collapse. Nasir points to Pakistan, one of the world's biggest freelance markets, recording $1.76 billion in freelance foreign exchange earnings, up 78% year-over-year, per State Bank of Pakistan data. Upwork reports freelancers working on AI-related projects earn roughly 44% more per hour than everyone else on the platform.

The tasks that were pure commodity execution—logos, product descriptions, basic code—are getting priced out fast. The freelancers who learned to direct and refine AI output instead of competing with it are getting paid a premium. Nobody had to ban anything for that repricing to happen. It happened because buyers stopped paying for work a machine does cheaper.

ADP Research, working with Stanford's Digital Economy Lab, found a more specific pattern inside companies: AI is best at structured, repeatable work, which historically was junior-level work, according to Fast Company. The risk isn't senior people getting replaced. Entry-level jobs—the ones that used to teach people judgment and pattern recognition—are the ones getting automated away first. That's a real workforce pipeline problem, and it applies whether you're running a law firm or a marketing agency.

What This Doesn't Prove

None of this is evidence that gig work itself was a bad idea, or that platforms are cheating people by definition. Millions of workers, from Pakistani freelancers to Edinburgh cyclists, chose flexible platform work over nothing, and for years it paid better than the alternative. The Straits Times is right that gig platforms became a safety net for people who'd fallen out of the formal labor market. What's changed is that the safety net now has two holes in it: opaque algorithms squeezing physical gig work, and AI hollowing out desk-based freelance work.

The class action against Uber is headed toward litigation, not a settled fact. The Workers' Observatory's decade of funded research will produce more data on how dynamic pricing actually works, which is the kind of transparency that could settle the argument either way. Until then, riders like David are working the same hours for less money, and nobody outside the platforms can say exactly why.