No, AI is not ripping off Uber drivers
Developments in artificial intelligence and machine learning have led to a series of debates over, among other things, whether it’s coming for your skilled job. AI anxiety has now crept into the domain of the gig economy, but not in the way you would expect.
Recent articles point the finger at companies allegedly using AI to “get inside gig workers’ heads” to “manipulate” their pay. The fury was jumpstarted by a law review article claiming that companies are engaging in “algorithmic wage discrimination,” a fancy term that means ridesharing drivers may be offered different rates for what seems like the same type of work.
It’s not unreasonable to think AI will make ridesharing apps better at figuring out how to pay drivers. That doesn’t necessarily mean drivers will earn less or that the rates they are offered will be determined in unethical ways.
In the consumer setting, travelers should already be well-acquainted with the concept. If you looked at prices for a flight six months ago, would you be surprised to find that it costs three times as much a week before departure? With fewer seats available, an airline can allocate them either on a first-come, first-served basis or through the current system. The current system is both more profitable for them and more likely to ensure that the person who needs the seat most can get it.
This dynamic pricing mechanism, which has been used by airlines and hotels for decades, is also in play in the ridesharing industry. Prices for your Uber ride “surge” when you try to book during rush hour or after a major event in your area. What’s more, if you take an Uber from your house to your office at the same time, every day, you will still see different prices for that same trip. That’s because the pricing algorithm used by Uber and other ridesharing companies takes into account riders’ demand and the supply of local drivers at that exact time, which, of course, varies from minute to minute and from day to day.
The different price that the consumer sees is also why the driver earns a different rate for what might seem like the same trip. But these are in fact distinctive trips in that the current market conditions are never exactly the same. For example, the trip from a bar scene to a residential area, even if it takes the same amount of driving time, will yield different prices (and therefore different rates for drivers) at 2:00 in the afternoon than it would after a 2:00 a.m. closing time.
That’s not necessarily a bad thing. It costs more to entice drivers to go out late at night, into inherently less safe conditions, and pick up intoxicated customers. Indeed, it would likely be unfair to pay drivers the same rate for daytime and closing bar trips. In fact, female Uber drivers’ understandable reluctance to work during the twilight hours partly explains the differences in their take-home pay with male drivers.
If regulations prohibited different rates for those very different jobs, then we wouldn’t see nearly as many Uber drivers out in the middle of the night. Why take the undesirable night shift if it pays the same as the more desirable day shift?
Of course, some drivers may prefer more predictable rates. The law article on “algorithmic wage discrimination” provides interviews with some Uber and Lyft drivers who raised real concerns about how unpredictable rates and “fake price surges” negatively affected them. While these experiences are genuine, they are not generalizable to the whole population of drivers. The interviews came from self-organizing drivers and those at protests precisely because they were already unhappy about their experiences. It would be akin to asking Trump supporters at a rally whether they liked the Biden administration’s policies.
In other words, some drivers see unpredictable rates as a negative part of their job experience, whereas others like that they can choose to work during busy or more lucrative times. Indeed, economists have found that drivers extended their sessions and provided significantly more rides to more people when they saw “surge pricing” in a particular geographic area.
That’s not to say that companies can’t possibly be engaging in other unethical practices behind the scenes to determine how rates are calculated for individual drivers. But currently, there is no real evidence of AI manipulation beyond what is seen by everyone — drivers get different rates per ride because riders pay different prices per ride.
Liya Palagashvili is a senior research fellow with the Mercatus Center at George Mason University.