The Power of Algorithms in Online Ride-Hailing Platforms

Algorithms and artificial intelligence are becoming the weapons of choice for online taxi and ride-hailing companies in the future. What does this mean for the fate of their drivers?

FOR the past ten months, Firman has been sleeping inside the car he is still paying off in installments. Having left his hometown of Palembang in South Sumatra for the capital, Firman works as a taxi driver for app-based platforms such as Grab Car or GoCar. Sleeping in his car allows him to avoid paying rent or boarding-house, so he can continue sending money home to his wife and three school-age children. 

Firman works up to 18 hours a day. At 11 p.m., he switches off the app and heads to the closest 24-hour gas station where he can park the car and get some rest. By 05.00 the next day, he wakes up, turns the app back on, and waits for orders to come in.  

Before becoming an online taxi driver, Firman worked as a dump truck driver for an oil and gas company in South Sumatra. In October 2015, he was involved in an accident while on duty. His left arm was broken and he could no longer drive large trucks.

---

As a nonprofit journalism organization, we depend on your support to fund more than 170 reporting projects every year on critical global and local issues. Donate any amount today to become a Pulitzer Center Champion and receive exclusive benefits!

---

With one good arm left, Firman resigned from the company and began driving an online taxi in Palembang. “Eventually, I moved to Jakarta because there weren’t enough orders in Palembang,” he recalled.  

To keep up with his car payments, survive in Jakarta, and send money home, Firman needs to earn Rp 250,000 a day, or approximately Rp 7.5 million a month. But in recent months, according to Firman, this target has been increasingly difficult to meet. 

To make matters worse, Firman’s account on the app is frequently suspended for rejecting orders that he considers unprofitable. He feels punished by the platform. “There’s no way I can take low fares. I usually only accept orders above Rp 50,000,” he said. 

The way online taxi drivers respond to incoming orders varies widely, and is shaped by traffic conditions, physical health, driving skills, habits, and income targets. However, the algorithms used by digital transport platforms do not take these factors into account. Instead, algorithms have taken on the role of an HR manager, readily imposing sanctions on workers. As a result, drivers are left with increasingly little room for negotiation. 

According to Ayom Purbandani, a researcher at the Center for Digital Society (CfDS) at Gadjah Mada University, digital platform algorithms typically serve four key functions, including setting fares, optimizing performance, and imposing sanctions in the form of automated suspensions. 

But, Ayom added, these systems have also replaced the role of human managers. Shifting the managerial authority over drivers to algorithms and artificial intelligence (AI) allows platform companies to distance themselves from responsibility. “AI becomes the invisible boss, further obscuring the role of platform owners. Platforms who hold significant power find it even easier to wash their hands of accountability,” said Ayom. 

Meanwhile, Tidar University lecturer Arif Novianto argues that algorithms are deliberately designed to extract labor from online taxi and ojek (motorcycle taxi) drivers. “The goal is to discipline them,” said the author of Cheap Labor Regime in Platform Capitalism. 

Brutal Hours, Meager Returns 

The combination of algorithmic control and low fares forces most online taxi drivers to work all day for meager returns. That is the conclusion of a survey conducted by the reporting team among 40 Grab and Gojek driver-partners. 

The survey shows that the earnings of online taxi and ojek drivers this year have fallen to levels close to those seen during the Covid-19 pandemic. In October 2025, the average income of 20 online ojek drivers was Rp 112,250 a day, while the average daily income of 20 online taxi drivers was Rp 271,464. 

These daily income figures fall below the findings of a 2023 study by the Institute for Demographic and Poverty Studies (Ideas), which recorded an average of Rp 174,805. The reporting team’s survey indicates that incomes are again approaching pandemic-era levels, or periods of low passenger demand, at around Rp 100,000 per day. Before the pandemic, average daily earnings could reach Rp 304,688.  

Sixty percent of respondents said they work 10 and 14 hours a day, and 57.5 percent said they work seven days a week. Meanwhile, 21 respondents are still paying instalments on, or renting, the vehicles they use for work. Instalments or rental fees make up around 50 percent of their gross earnings, excluding fuel costs. 

Despite having endured these working conditions for years, online taxi and ojek drivers remain divided over what kind of regulation could secure their welfare. Some want to be recognized as formal workers, which would grant access to labour protections such as insurance and income stability. Others are willing to retain their status as “partners,” fearing that platform owners would otherwise cut jobs to reduce costs.

Some drivers support a 90:10 revenue-sharing arrangement, while others oppose the ratio, worried that the loss of customer promotions and incentives would lead to fewer orders. They also fear that allowance for vehicle maintenance, basic necessities, and data packages would be removed.

On the other hand, said Arif Novianto, the government tends to endorse platform company policies, particularly on the issue of drivers’ status as partners and the revenue-sharing scheme. Conditions on the ground show that, in addition to taking a 20 percent cut from the fares paid by passengers, platform companies impose additional charges on users, including “Platform Fees,” “Ride Cover,” “Priority Booking,” “Faster Search,” “Safe Trip Fee,” among others. These additional fees go entirely into the pockets of the platform companies.  

When these fees are taken into account, platform companies are effectively taking more than 20 percent of the total passenger fares for their role as intermediaries. This arrangement allows them to avoid formally breaching government regulations. 

If given the choice, many drivers interviewed by the reporting team prefer to have another formal job. For them, driving for ride-hailing platforms is a last resort for earning an income while no other work is available. 

Formerly a supervisor at an automotive company, Firmansyah has been working as an online taxi driver for a year and a half. To the author, he admitted that he preferred his previous job. Firmansyah complained about the long working hours and fierce competition among drivers to secure orders. “Well, this is because there are so many unemployed people in Indonesia,” he said. 

Firmansyah works without a single day off. He starts at 6 a.m. and stops working at nine in the evening. “The [car] rental is not going to pay for itself. I just hope I don’t get sick,” he said. 

In addition to employment status and fares, Taha Syafariel, chair of the Online Drivers’ Association, said that the government must also regulate the algorithms used by platform companies. “There needs to be transparency, including clear notifications about when drivers are sanctioned and the reasons why,” he said. 

According to Taha, platform algorithms often feel like a puzzle. Drivers struggle to understand why the system suddenly assigns them many or few orders, why fares are raised or lowered, or how it sends short- or long-distance trips. He added that drivers are often sanctioned suddenly, without clear explanation. 

Meanwhile, Raymond Kusnadi, coordinator of the Indonesian Transport Workers’ Union, said that algorithms carry an element of command: “If those commands are not followed, the sanctions can range from suspension to termination of partnership.” He added that platform companies recruit far too many drivers, leaving them unable to secure enough daily orders to meet their income targets. “We argue that this is a form of technology-enabled modern slavery.” 

The Arrival of Generative AI

Amid these mounting problems, digital companies have begun promoting generative AI as a solution that promises to help online taxi and ojek drivers increase their earnings. The technology is framed as the next phase of digital evolution, one that guarantees benefits for everyone. 

One example is “Sahabat AI,” a chatbot developed by GoTo in partnership with telecommunications company Indosat and accessed through the GoPay application. Meanwhile, its competitor, Grab, has teamed up with US-based AI developers such as OpenAI and Anthropic. 

Grab is working on a feature known as AI Driver Companion. The tool allows Grab’s driver-partners to submit real-time data on road conditions using buttons and voice input. Through this feature, GrabMap, the navigation system used by drivers on this platform, can be updated continuously and with greater accuracy. 

“With this feature, we can increase driver productivity, as measured by the proportion of time spent online completing tasks rather than waiting idly,” said Anandhita Kasetra, Grab’s Head of Product Strategy and Product Operations, in an interview with the author. 

Under the supervision of Grab’s communications staff, we interviewed two Grab Car driver-partners, Ricky and Bima. Both use the AI feature and said it has helped boost their productivity and earnings. “It helps us figure out which roads aren’t congested. That matters, because time is money for us,” said Bima. 

Ricky and Bima also agree that GrabMap has become more accurate thanks to the new AI feature. They both work seven days a week, without any days off. 

Technology companies have relied on AI in their algorithms to manage a wide range of activities. Social media companies, for example, use AI to curate information; Netflix and Spotify to recommend content; ride-hailing platforms use it to regulate fares and driver performance, and e-commerce platforms use AI to control the visibility of merchants. 

The explosive growth of generative AI has given digital companies fresh momentum. Data extraction has become easier, paving the way for new corporate visions and ideas about automation. 

For example, Grab envisions a future of mobility that relies not on public transportation nor private vehicle ownership, but rather on driverless rental vehicles. The vision is echoed by executives at Uber, who describe it as “mobility as a service.” 

According to Grab co-founder Anthony Tan, such vehicles could be fully operated by computers or controlled remotely by human drivers. Online drivers, he said, are expected to upskill and transition into roles as remote drivers, data labelers, camera and sensor operators, vehicle maintenance workers, and other support functions. 

In line with this vision, Grab began investing dozens to hundreds of millions of dollars in various startups developing autonomous vehicle technology, like May Mobility, Vay Technologies, and WeRide. Grab is confident in its capacity to develop driverless vehicle technology thanks to a valuable asset already in their possession: GrabMap. These autonomous vehicles will be trained on GrabMap data, allowing them to operate with little to no human assistance. 

As Ricky and Bima have testified, GrabMap’s accuracy is built and continuously refined by millions of Grab drivers today. This is done through activities like verifying the names of places, driving through the streets, and reporting traffic or road conditions. The AI Driver Companion feature makes these tasks easier. In other words, the labor of low-income drivers who work without days off directly contributes to the development and training of this future technology. 

Without transparency, drivers risk being steered not toward their own interests, but toward greater business efficiency for the company, said Ideas researcher Muhammad Anwar. “In situations like this, AI assistants shift from being tools of support into instruments of control, boosting productivity while narrowing drivers’ room for autonomy,” he added. 

Anwar is skeptical that such innovation will be able to absorb large numbers of workers. “Even if Grab promises reskilling programs for drivers, not everyone will be able to adapt or be absorbed seamlessly into this new ecosystem,” he said. 

Driverless or autonomous vehicles (AV) have already entered the mainstream imagination in China and the United States. In Southeast Asia, however, their development is expected to advance more slowly, largely because the cost of human drivers remains lower than that of automated vehicles. Companies like Grab will naturally choose the cheapest option. “It will take quite some time before the economic unit cost of driverless vehicles can match that of human drivers,” said Anthony Tan. 

This means that once the cost of human labor rises to match or exceed the cost of automated vehicles, online drivers may face a new challenge almost immediately: competing for work against AI-powered driverless cars and remote drivers. 

For now, millions of online taxi and ojek drivers continue to compete with one another on the streets. Armed with a smartphone and an app, they flock to areas labelled “high demand” in the hope of catching passengers. They sleep by the roadside, at electric vehicle charging stations, in car parks, at roadside stalls, or at gas stations. “I hope the app won’t act arbitrarily toward its partners, and will respect the sacrifices we make,” Firman told the author, four hours into searching for passengers in Jakarta’s crowded Senen area. “I also hope there will be a legal framework to protect online drivers.”