The Decibel - When AI decides your pay

Episode Date: July 13, 2023

App-based ride-share and food delivery companies rely on algorithms that dictate speed, behaviour and the wage of gig workers resulting in different payments for the same work. Labour experts are warning about ‘algorithm wage discrimination’ and concern around protecting workers when it comes to AI.Vanmala Subramaniam, The Globe’s future of work reporter, discusses what we do know about how this works, and the growing trend of AI being in charge of a worker’s pay.Questions? Comments? Ideas? Email us at [e-mail]

Okay, so this is the McDonald's. So you would just wait behind this white car, presumably, and run in and get it? Yeah, yeah. The Globe's future of work reporter, Vanmala Subramaniam, is in the car with Jude Okoye. He's working for DoorDash, the grocery and food delivery app. You have to be a very ace driver to do this job. I would never be able to. I don't do superfood ever.

See what I'm going to do? Okay, now I click arrive, right? Yep. No, it's fine. I'm going to wait. Indy arrived. Vanmala wanted to get an understanding of how gig workers, like Jude, get paid.

The delivery apps they use are highly dependent on AI, and that means a worker's pay is decided by an algorithm. But how the algorithm decides remains a bit of a mystery. Today, Vanmala is on the show to explain what we do know about how this works, and why labor experts are worried about AI being in charge of a worker's pay. I'm Mainika Raman-Wilms, and this is The Decibel from The Globe and Mail. Van, thank you so much for joining me today. Thanks so much for having me, Mainika.

So you spent time with Jude Okoye while he was working for the DoorDash delivery app. Can I ask you, like, what was it like going around with him? Yeah, it was an incredibly interesting experience for me, someone who has never been a food delivery career. I think what I was most struck by was the pace at which Jude moved. He was constantly looking at the clock, constantly trying to cut his minutes down, doing mental calculations of if I get this done in three minutes, that slices two minutes of, you know, a full hour and I will presumably get paid more. That was kind of what's always on his

mind. But really, it was very impressive because it is not an easy job. These guys, you know, they start their days really early. They are sometimes working up to 12 hours a day, not consecutively, but, you know, they're out of the house for 12 hours. They are dealing with traffic, people living in condos with limited parking where you have to go up and deliver their food to the exact unit to guarantee yourself a good tip. You have to deal with restaurants that sometimes keep you waiting on the orders. even shadowing him. I truly felt quite exhausted after doing two four-hour shifts with him. Yeah. So, I mean, tell me about Jude, though. What did he say about why he does this work?

Yeah. So Jude's a very interesting person. He's Nigerian. He immigrated to Canada back in the late 90s, spent much of his time working in the warehousing industry. In 2018, he started using delivery apps. He basically told me that he really enjoys the flexibility of the job. Although, as we'll see, there are a lot of difficulties with the job, especially when it comes to how Jude is paid. Let's talk about this, how these apps work then from their side, because I think a lot of us know how they work from the user side, right? If we're ordering Ubers or food, we open the app, we choose what we want. It just shows up at our door, really, right? It's as easy as that.

But let's break it down from the worker side here. So Jude is working for DoorDash. So how does the process work for him on the app? All right. When Jude starts his shift, he parks at a certain part of Toronto where he wants to start delivering in. He logs into his app. The app then asks him, what zone do you want to deliver in? He clicks on Midtown Toronto.

Then he waits. And this is what is really interesting. When I shadowed Jude, we started at 11 a.m. He waited 20 minutes until he got the first order generated to him. So keep in mind that the drivers, the delivery couriers have no control over getting an order. They are sitting in their car waiting for the app to send them an alert, which then says, you have received an order. Would you like to accept? And then the order says the distance that it'll take, the pickup address of the food, so a McDonald's nearby, for example, a sushi place.

It does not give Jude the exact drop-off location, but it gives Jude the estimate of the distance it'll take for the entire sort of order. If you accept the order, it says you will be guaranteed an approximate amount of $9, $10 for the whole order. So you know basically, all right, I'm going to make about $10 for this delivery. And then you start doing the delivery. Are those numbers accurate? Like for the distance and for the money calculation there, is it generally what a driver would end up getting? It's generally accurate. That is what I found.

The inaccuracy is that the amount you get, say $10, doesn't correspond to the time you spend doing the delivery or the distance you drive. So there really isn't a sense of if I work and do this many deliveries in this amount of time, I should make, say, this amount of money every day. That varies. And that's what confuses the drivers, because sometimes they just can't tell you know how long it'll take in traffic say to reach a certain spot they might have spent 45 minutes going through traffic and they'll get the same amount as they do from doing a delivery where they take eight minutes so this is exactly what drivers and delivery couriers are saying is the problem with the app, with these companies.

And, you know, your wages are just so uncertain. And there's sort of an information asymmetry between what the company knows and what the driver knows. I also was have been in contact for a long while with another Uber Eats, you know, delivery career uses a bike in downtown Toronto. And, you know, it's kind of the same problem. Time distance doesn't really correspond to wage. Okay, let's let's get into how a worker's actually compensated here, Van. How is their pay calculated? With DoorDash, a worker's pay is calculated as a combination of three things. Their base pay, the tips that they get from customers, and what they call promotions. And Uber Eats calls it a boost or a surge. Now, promotions,

boosts, surges, they are monetary incentives of about about one to three dollars per order that's given to the worker to deliver in a very busy neighborhood or at a very busy hour. Tips and promotions are variable, so you never know what exactly you're going to get. The base pay, according to DoorDash ranges per order from $2 to $10. Now, when I followed Jude for almost all of his orders, I would say 99%, he never got a base pay of more than $5. Okay, so it's determined on those three factors. But I guess like, who's making the decision based on those factors about how that person gets paid? So we do know that these companies don't have actual human beings behind the scenes generating each order for delivery workers all over the world. That would be quite impossible to do. Nobody's frantically doing calculations in the back room or anything. Yeah, exactly.

So we know that this is technology doing it. We know that it is an element of artificial intelligence that is an algorithm programmed in that is based on a myriad of factors that determines what workers' ultimate payout is. Now, the thing we also don't know is how do these algorithms change over time? How do they change according to geography, according to time, according to who is doing the delivery? All of these things are completely unknown. There is no transparency from the companies about each transaction and each order. So experts say that what they do know is that these are algorithmic wages. That's a term that they've come up with. And what they do know is that the algorithm changes. That's about it. We don't know exactly how these companies really

program the algorithms. We'll be right back. What did Jude tell you about this? Like, he's been doing this for a while. Has his base pay kind of, I guess, always stayed similar to the range that you saw? Yeah, so Jude said when he first started using DoorDash, and this was maybe late 2018, early 2019, on a single day, let's say he worked two four-hour shifts with a break in between, he managed to take home about $250. And that's over. That's above minimum wage in Ontario. Now, this obviously doesn't include taxes he would have to pay on his wage, his cost of his gas and food that he has to eat during the day. But he claims that over time, especially in the last one year, he has noticed that he can never make as much as $250 a day, even though he does the same amount of deliveries. So he finds now that he makes

slightly maybe under $200 a day when he does two four-hour shifts. So at the end of the day, my calculations showed me that his net income after costs was $185, and that was for a 12-hour day. And so that's, I mean, that's anecdotal. That's what he's seeing. Are we seeing any research around this, though, Van, like that, I guess, can broaden this out a little bit? Yeah. So, in fact, there is research that proves Jude's point. So in late 2022, New York City's Department of Consumer and Worker

Protection found that food delivery workers on five different apps, Uber Eats, Grubhub, DoorDash, Relay, and Skip the Dishes, earned an overall pay that was about $14 US dollars per hour. Now, within that amount, they found that the base pay per delivery had declined by 23% between the first quarter of 2021 and the second quarter of 2022. So they have found that base pay is declining in these apps for whatever reason.

And they also found that workers, because the base pay is declining in these apps for whatever reason. And they also found that workers, because the base pay is declining, they're increasingly dependent on the variable aspects of compensation, which is the promotions they get and mostly tips. How much do these workers actually rely on tips then? Yeah, so what's fascinating is in this entire excursion I did with Jude, he made more than 50% of his wage from tips. More than half. Wow. Exactly.

And he says that is standard. So really the app companies only compensate him for half of the amount that he earned on that day. The rest of it comes from the generosity of, you know, customers. But he's always worried that that trend will shift and that people will say, look, I don't think I want to tip 18 to 20 percent anymore. This is getting ridiculous. And go back to tipping him 10 percent, which would severely affect his earnings. So what have what have the companies said? What have companies like DoorDash said about how they calculate pay for workers?

DoorDash sent me a statement where they said their pay model is designed to make earnings fair and transparent for every delivery. And they said that they do provide dashers, so their food delivery careers, with a clear understanding of potential earnings on each order, and more importantly, the flexibility to choose how they deliver. They also said, gave me a numerical figure saying that, on average last year,

door dashers earned over $27 per hour while on delivery. And that included tips. But they also say that a lot of their workers do this part time. On average, they deliver fewer than three hours a week. The point they were trying to make here is that the drivers do have full knowledge of what they're going to get paid before they accept a delivery. And that pay model is really designed to be fair and transparent. And what do we know about the revenue of the companies behind these apps? Are they making a lot of money on this? I found that Uber Eats in particular saw a revenue surge of about 23% between the first quarter of last year, 2022, and the first quarter of 2023. 23% is a substantial increase.

Now, DoorDash, the overall revenue, again, from last March, 2022 2022 to this March, increased by 40%. So these companies are starting to make a lot of money. It took a lot of time for them to break even, but they are generating more revenue than they used to. Yeah. Workers who are delivering for these companies are not making any kind of minimum wage, right? Usually there's provisions in place for other workers, but not necessarily the case here. I wonder, though, because we often look to government to regulate these kinds of things, are there any government policies that the labor experts, you were speaking to, Van, that they think should be considered to actually help make these wages more equitable for these workers? Yes. So there are a couple of

legislative battles going around in North America that's pushing for workers to essentially just get paid a minimum wage per hour, which would really give them some sort of consistency and predictability to what they're going to earn. Now, in Ontario, there is a bill that's been passed and that basically calls for delivery workers and Vrija drivers to be paid a minimum wage, but the caveat being only for the time they spend doing the delivery and not for the time that they spend idle kind of waiting for a delivery. And labor advocates say that doesn't do the trick because that essentially doesn't compensate them for time that where they're on the job, but through no fault of their own, they're just waiting for orders.

More interestingly, in New York City last month, they instituted a minimum pay standard that was very generous of $17.96 US dollars per hour. Doesn't matter if you're idle in that hour. And, you know, that amount was considered a win for gig workers. Yeah. It's interesting because we're hearing about AI being used in so many different ways now, algorithms being used in so many different ways as well. Based on your reporting, I wonder, could other sectors be adapting this kind of use of algorithms and artificial intelligence essentially and how they compensate workers? Because it sounds like here, I mean, these gig workers are basically getting paid whatever the algorithm thinks they should be getting paid. There are workers in the U.S. in certain hospitals, nurses who get compensated based on algorithms for the kind of work that they do, the procedures that they do, a combination of that and the shift, the length of time that they work. So there are little examples popping up here and there where your base pay is being decided by a machine and not by a human being. The gig worker rideshare driver experiment is extremely interesting for people who study wages because

what they say is it essentially undermines a lot of things that labor advocates have fought for for decades. Right. You know, minimum a minimum pay standard rights that workers have that allow them to push back against an actual person. If they feel that there's no pay transparency or if they're being you know, they're doing the same work and they're being compensated differently for it. You can deal with the person. Now, if an algorithm makes these decisions and there's very little transparency around how exactly those decisions are made, it opens up sort of a Pandora's box of how exactly are we going to decide who gets paid what. Just very lastly here, Van, I wonder, how does Jude feel about how he's being treated in this line of work? He feels, on the one hand, grateful that he has an option of making money in a way where he doesn't really have to apply for a job and go

through an interview process and deal with people. So he likes that aspect of it and the flexibility. What he feels very upset about and frustrated about is that he just doesn't feel that he's getting paid enough for what he does. And when I specifically asked him, what would you want to see yourself make? He says that he just wants the base pay per delivery to increase. And if that happens, he has more predictability about what he's going to get paid per order, which then allows him to plan his day better. But when I asked him, do you feel like the solution is you becoming designated as an employee? He says, sure, as long as that doesn't take away the flexibility of me doing the job.

He doesn't want to be wedded to an employer for a full eight hour day. He wants to be able to log in and off the app. So it's a bit complicated, but fundamentally, Jude feels that he's being underpaid. Van, thank you so much for your reporting here and for taking the time to speak with me. Thank you so much for having me. That's it for today. I'm Maina Karaman-Wellms. Our summer producer is Nagin Nia. Our producers are Madeline White, Cheryl Sutherland, and Rachel Levy-McLaughlin.

David Crosby edits the show. Adrian Chung is our senior producer, and Angela Pachenza is our executive editor. Thanks so much for listening, and I'll talk to you tomorrow.