At the end of 2018, when it was a little less than 10 years old, Uber announced that each month more than 91 million people were using its services to order taxis or takeaways around the globe. It’s a remarkable number. But a second number also stood out: that the company could count on 3.9 million drivers. Not a single one of these drivers worked for Uber. They were “gig workers.” They didn’t have an employment contract with the firm but, rather, were paid, in a roundabout way, for every customer they drove or every freelance gig they completed.

Uber offers a neat demonstration of how we get work in the exponential age wrong. For all the talk of mass automation — talk that Uber, with its much-hyped self-driving car program, is all too keen to promote — the company has created millions of jobs. But there is something unusual about these jobs. Uber is one of the biggest companies to use networks of freelancers, rather than contracted employees, for its primary business operations. It isn’t a small company by any means — it has more than 20,000 employees, none of whom are drivers. Yet for every full-time employee, there are nearly 200 drivers working anything from a few hours a week to 10 hours or more a day. Uber has demonstrated that platform-based gig work can function at an enormous scale. These new working arrangements, rather than automation, raise the trickiest questions relating to employment in the exponential age.

While Uber is probably the most successful platform-based freelance work company, it did not pioneer the concept. The origins of the gig economy — where short-term, freelance tasks are allocated by an online service — lie in Amazon’s Mechanical Turk platform, launched in 2005, a few years before the term “gig working” was coined. The service gets its odd name from a famous chess-playing device of the late 18th century. Through the 1770s, the Mechanical Turk — a mannequin affixed to a chessboard that was mounted on a wooden crate — made waves by beating successive royals, aristocrats, and statesmen at chess. Nominally, it was powered by an ingenious machine inside the box. In fact, the Turk was operated by a human: a chess master who crouched inside the crate and manually moved the pieces.

Like the original Mechanical Turk, Amazon’s version offers up an apparently automatic way to get things done. And like the original, it is actually underpinned by hidden human labor. The interface is similar to the odd-jobs notices you sometimes find posted on notice boards at the back of neighborhood stores. Tasks on Mechanical Turk tend to be quite small and well-defined, but just beyond the reach of current AI systems. And so humans have to step up. A typical job, called a Human Intelligence Task, might be to go through a list of company websites, find the addresses of their branches, and copy them into a database. Anyone can apply to undertake these tasks, and be paid a small fee for every one they complete. Within a couple of years of its launch, more than 100,000 Turkers, as workers on Mechanical Turk are known, had registered with the service. 

At first, the tasks on Mechanical Turk had a particularly techie flavor. More than 19 out of 20 of the jobs on Mechanical Turk related to getting information about digital images or collecting information from other websites, each task being worth about 20–30 cents to the Turker. The workers on Mechanical Turk became an incredible ally to companies dealing with large volumes of data. Many of the amazing machine learning systems that emerged during the late 2010s were capable because of the mind-numbing work of thousands of humans manually classifying data for the algorithms to learn from.

In time, this type of activity garnered a new name: crowdsourcing. The internet could connect people who needed something done with thousands, perhaps millions, of those with the time and skill to do it. According to Jeff Howe, the professor of journalism who coined the term, crowdsourcing would unleash “the latent talent of the crowd.” In these early days, the notion of crowdsourcing had a utopian feel: millions of people working together, perhaps voluntarily, to build some incredible tool like Wikipedia.

Within a few years, crowdsourcing platforms had multiplied. Services like Elance and Odesk sprung up for complicated tasks, like programming or copywriting; Fiverr and PeoplePerHour were created for tasks that were less complex than programming but more complex than an Amazon Human Intelligence Task. And if the internet got the trend started, the smartphone helped it take off. The phone became omnipresent. In-built global positioning systems meant phones always knew where they were — allowing crowdsourcing platforms to offer us local, highly convenient services. Soon, we could order taxis, takeaway food, and massages from the comfort of our couches. TaskRabbit, now owned by furniture giant Ikea, will dispatch someone to help you assemble your new bookcase. Talkspace will help you find a therapist. Wag will find a walker for your dog. In time, the term crowdsourcing, which often referred to unpaid, noncommercial work, gave way to a new term: the gig economy.

This whole new way of working was underpinned by the emerging exponential economy. Crowdsourcing depended on digital platforms. These platforms were able to scale because they were susceptible to network effects and had access to an exponentially increasing amount of computing power. The tasks — sifting through data, finessing product ideas — invariably related to the development of intangible assets. And it was all facilitated by two key general-purpose technologies of our age: first the internet, then the smartphone.

Ten years in, and gig economy platforms are continuing to grow exponentially. They have upended once-stable markets. By 2017, a mere six years after entering New York, Uber’s drivers ferried more passengers than yellow taxis. In the U.S., Uber matches more than one million rides every single day. Such successes have turned into revenues, in 2019, of $14 billion. And that platform growth has meant more gig workers. In the U.K., in 2019, 2.8 million people were estimated to be platform workers, a shade under 10% of those considered employed. Globally, digital platforms could add the equivalent of 72 million full-time positions to the global labor market by 2025. Within two decades of the launch of Mechanical Turk, digital piecework might have increased the international workforce by as much as 2%.

All this points to a truth that the hullabaloo about the “robopocalypse” doesn’t capture. Gig work is a more imminent and transformative force than mass automation. But what does it actually mean for workers? Evangelists for the gig economy — among them, naturally, the founders of leading gig companies — say that their model helps workers in two key ways: It can make markets bigger and more efficient, creating more opportunities for workers, and it can improve the quality of work that someone does. 

If gig work is generally more flexible and less formal in richer countries, the reverse is true in poorer ones.

Many labor markets are inefficient. Demand for a particular type of work goes unmet, perhaps because it is hard for employers to find workers or vice versa. Or perhaps there are middlemen taking an unfairly large piece of the pie. Gig working platforms make it easier to connect buyers and sellers; the databases and algorithms do the matching. This can lead to an overall increase in opportunities for workers. In developed economies, Uber is bigger than the taxi businesses in many big cities, which is evidence that the company is making markets bigger. In emerging economies, labor markets are often clunkier. Kobo, a kind of Uber for freight, has helped Nigerian truckers get work in a famously inefficient market mired by corruption and bureaucracy.

If gig work is generally more flexible and less formal in richer countries, the reverse is true in poorer ones. In emerging economies, a gig working platform may offer more security, more employment options, and greater freedoms than casual or day labor. In India, for example, the sheer size of the informal labor market gets in the way of the government’s ability to spend on health and education. Casual laborers, hired daily, paid in cash, rarely pay income taxes. Nor do their employers contribute to payroll taxes. Lower tax participation means less booty in government coffers to fund social programs. For highly casual labor markets, the gig economy could be a route to a large, more formal sector with more protections for workers and a more robust tax base for governments.

So far so good. Yet, there is evidence that all is not rosy for workers on the digital labor platforms, especially in advanced economies. Pay is often poor compared to traditional work; working patterns can be precarious, offering few protections, should a worker get sick. The platforms themselves can make unilateral changes to how they operate and what they pay. Many companies maintain internal scores for the workers on their platforms, which might affect what jobs they are offered. And equally, unions or other collective arrangements are uncommon among the independent workers on gig platforms, which means there is often no collective voice to represent their interests. What this boils down to is a huge imbalance in bargaining power between the platforms and their armies of labor. 

As these smartphone-based gig companies sprung up in the mid-2010s, they also led to increasingly precarious working conditions. The new tech platforms go to great pains to explain that the people doing the work are not their employees. The platform’s job was merely to introduce you to someone who would drive you to work or deliver you some late-night ice cream. They were like a modern-day temping agency. 

Such distinctions matter. In many countries, particularly in the developed world, employees are treated markedly differently than the self-employed. Employment involves a clear trade-off. The employee enjoys a bundle of benefits: a regular wage, job stability, and hard-won labor rights: the right to an annual wage, sick pay, parental leave, and to a fair dismissal. In return, the employer gets the time and best efforts of the employee. The deal is stability in exchange for subordination. The self-employed, on the other hand, have always been at the mercy of the market. Work could be irregular, and sickness meant days off work with no pay. 

Companies using gig labor, the largest of which are in the ride-hailing and food delivery sectors, have generally been reluctant to strengthen the protections they offer their workforces. These problems are down to an exponential gap between new modes of employment enabled by exponential technologies and a set of labor laws designed in the 20th century. The great triumphs of the 20th-century labor movement were about securing humane working conditions for contracted employees. The eight-hour day, sick leave, pensions, and collective bargaining were all extended to those who were formally employed by a company. But in the exponential age, formal employees are relatively decreasing in number. The technologies of the exponential age create new ways of working, with the smartphone and the task-matching algorithm allowing firms to rely on pools of freelance talent. And our labor laws haven’t yet caught up. That workers are forced to rely on court decisions rather than clear rules reveals that the gap has not yet been closed. 

All of this leads to growing inequality between gig workers and official employees. The self-employed have always been at the mercy of the market. But in the exponential age, their number could swell to the hundreds of millions. Only a small group will retain the privileges that workers fought to gain over the last 150 years.

This piece is excerpted from The Exponential Age: How Accelerating Technology is Transforming Business, Politics and Society by Azeem Azhar. Copyright © 2021 Azeem Azhar. Printed with permission of the publishers, Random House Business and Diversion Books. All rights reserved.