Gig Work Ethics
The Algorithm That Decides If You Eat Tonight
When your boss is a piece of software, there's no HR department, no appeals process, and no one to look in the eye when things go wrong.
The Idea
Gig platforms have pulled off a remarkable sleight of hand: they've created a workforce of hundreds of millions without, technically, employing any of them. The legal category of 'independent contractor' did exist before Uber or DoorDash, but it was designed for consultants and freelancers who genuinely set their own terms. What's new is algorithmic management — the use of automated systems to direct, evaluate, and discipline workers in real time, with all the control of employment and none of the obligations. This creates an ethics problem that is easy to miss if you're focused on the classification debate alone. The deeper issue isn't just whether a courier deserves a minimum wage or sick pay (though they do). It's about accountability in systems where no human being is making the consequential decisions. When a driver is deactivated — the platform euphemism for fired — the decision is often made by an algorithm interpreting ratings, cancellation rates, or fraud signals. The worker receives a notification. There's no manager who reviewed the case, no one who weighed the context. The code just decided. This opacity is ethically significant in a specific way: it severs the relationship between power and responsibility. Someone built the algorithm, someone set its parameters, someone owns the platform — and yet the system is designed so that no individual is ever clearly accountable for an outcome that can devastate a person's livelihood overnight.
In the World
In 2021, a group of Amazon Flex drivers in the United States began comparing notes and discovered something troubling: they were being fired — 'deactivated' — based on AI analysis of their delivery photos. The system would flag a photo as suspicious, conclude that a package had not been properly delivered, and terminate the driver's account, sometimes within hours. The drivers had no idea the photo analysis was happening at all, let alone that it was the basis for termination. What made this case unusually clear-cut was what happened when the drivers managed to appeal in person: some were reinstated immediately once a human being looked at the same photos the algorithm had flagged. The AI had been wrong. It had misread shadows, angles, or reflections as signs of fraud. But the default was dismissal, not inquiry. Amazon is not unique here — it's just unusually well-documented. Across Uber, Lyft, Deliveroo, and Instacart, researchers and journalists have catalogued similar patterns: workers penalised for circumstances outside their control, rating systems that aggregate customer bias, and deactivation processes that can move faster than any human review could possibly keep up with. The platforms would argue this is efficiency. The workers would argue it's a system in which they bear all the risk and none of the recourse. Both are correct.
Why It Matters
This isn't just a labour story. It's a preview of a broader question we're only beginning to grapple with: what obligations come with deploying automated systems that make high-stakes decisions about people's lives? The gig economy is, in a sense, a testing ground. These platforms were among the first to use algorithmic management at scale, on a population with limited legal protections and limited bargaining power. The patterns they've normalised — opacity, speed, no right of meaningful appeal — are now spreading into other industries. Performance management systems in warehouses, scoring tools in lending, automated screening in hiring: the architecture is the same. If you work in any organisation that uses automated decision-making about people — and increasingly, most do — the ethical questions raised by gig work are your questions too. Who can a person talk to when the system gets it wrong? Who is accountable? What counts as a fair process when the decision-maker has no face? These aren't rhetorical. They're design choices, and they're being made right now, often by people who haven't thought of them as ethical choices at all.
A Question to Ponder
If an algorithm makes a decision that harms someone and no individual chose that specific outcome, who — if anyone — is morally responsible for it?
Get a new one of these every morning.
Start learning with Thinkable