The Impact of Robotaxis on Gig Worker Wages: Early Evidence and Future Outlook

This Week’s Focus: When Gig Wages Meet the Bot
Autonomous vehicles (AVs) are no longer experimental—they’re already operating in several U.S. states, and early data indicates declining earnings for human drivers. This raises questions about how automation, and GenAI, will reshape the broader gig economy. While causation has not yet been proven, tech leaders like Reid Hoffman predict gig-style work could become the norm by the 2030s. The challenge is clear: can we capture the efficiency of AI and AVs without eroding workers’ livelihoods.

Autonomous vehicles are no longer just experimental tech. They’re now actively operating in several U.S. rideshare markets.

Waymo rolled out its robotaxi services in Phoenix (2020), San Francisco (June 2024), Los Angeles (Nov 2024), and Austin (March 2025), while Tesla piloted “Cybercab” rides in Austin in mid-2025. These deployments offer a first glimpse at how driverless cars might affect human rideshare drivers’ earnings.

What can we learn about the impact on wages? Can we use the information to forecast the wider impact of AVs on the gig economy? And more broadly, can the results help us anticipate the fate of gig workers as they begin competing with Gen AI?

Let’s buckle up and see where the data takes us.

Rideshare Wages Under Pressure in AV-Active Cities

Early data from Gridwise Analytics (a firm I collaborate with on other projects) reveals a clear divergence: in cities where AVs operate, rideshare driver earnings have slipped, even as national trends improved. The median pay per trip increased +3.4% nationwide from July 2024 to July 2025, but in the AV hubs it decreased: in Austin by 5.3%, in San Francisco by 3.1%, in Phoenix by 2.4%, while in Los Angeles it remained roughly flat (+0.4%).

Hourly earnings tell a similar story. Across the U.S., median gross pay per hour in rideshare in July 2025 was about 1.0% higher than the previous year. But in every AV-active city, hourly pay decreased significantly. San Francisco saw the steepest plunge at -6.9% year-over-year, with Austin close behind at -5.3%. Los Angeles and Phoenix also saw hourly wages drop by roughly 4–5%.

In short, rideshare drivers in robotaxi cities are making less per ride and per hour of work than they did the year before, even as the typical U.S. driver’s earnings inched up. These declines in hourly pay signal mounting income pressure for gig drivers in AV markets. Notably, other large markets without robotaxi services largely followed the national pattern of stable or modestly rising pay, underscoring that something unusual is happening in the cities with AVs.

What’s behind the decrease in hourly earnings?

Hourly pay is a function of both per trip earnings and the total number of trips per hour. In San Francisco and LA, trips per hour dropped 5.1% and 9.9% respectively, versus a small 2.7% dip nationwide, which helps explain the weaker hourly pay. By contrast, in Austin and Phoenix, trip activity actually ticked up (trips/hour rose about +1% YoY in both). Drivers there stayed just as busy—or busier—which hints that in those cities, other factors like lower fares or shorter trips are driving down earnings. In other words, even when demand held steady, drivers in Austin and Phoenix saw per-trip payouts shrink enough to drag down their overall pay.

Indeed, lower compensation per ride appears to be a key factor. Gridwise found that driver utilization (time spent with a paying passenger vs. waiting) actually increased slightly in Austin (+1.3%) and Phoenix (+1.4%) over the year. Those drivers were occupied for more time, yet their earnings still fell—suggesting that pay cuts on each trip outweighed any gains from staying busier. In Los Angeles and San Francisco, utilization dropped (-8.6% and -3.1% respectively) beyond the modest nationwide dip (-2.9%). This implies a double whammy for LA/SF drivers—fewer trips and lower pay per trip—which translated into notable income declines.

Another factor is that rideshare companies are offering less incentive pay, which may or may not be related to AVs. Bonuses, quests, and other incentives per trip fell ~47% nationally year-over-year—a broad pullback in driver promotions, and something we’ve discussed before. However, in some AV cities the cuts were even deeper: Los Angeles saw incentive pay per trip plunge 65.3%, and in Phoenix it dipped 64.0% (Austin was a partial exception, with a smaller ~33% decline, better than the U.S. average).

These steep drops in bonuses are partly a nationwide trend—gig platforms have been tightening incentive spending across the board—but the extra-big cuts in LA and Phoenix suggest that an influx of AV capacity may be reducing the need for companies to entice drivers. Put simply, if robotaxis (plus existing drivers) already meet much of the rider demand, Uber/Lyft can offer fewer bonuses to attract human drivers, further squeezing take-home pay.

All these factors show up in the bottom-line earnings totals.

When we zoom out to monthly gross income, drivers in most AV cities saw sizable declines, even as drivers nationally earned more on average. From July 2024 to July 2025, median monthly driver earnings increased ~8% nationwide, but dropped 7.0% in Austin, 9.0% in Phoenix, and a hefty 18.4% in Los Angeles.

San Francisco was a curious outlier: monthly earnings there actually grew 7.8% —on par with the U.S. average. How did SF drivers manage higher total pay even as their hourly rate and trips per hour fell? The likely answer is they worked more hours. The report notes that San Francisco’s income growth “may come from drivers putting in more hours, since both hourly pay and trips per hour declined there.” In other words, SF drivers compensated for lower pay rates by grinding longer to keep their monthly income up—an interesting sign of how gig workers adapt when the market turns against them.

In summary, AV-active cities are showing clear signs of wage pressure on rideshare drivers, and this early pattern is consistent with an influx of driverless cars starting to displace some human-driven rides—or at least dilute driver opportunities.

However—and this is critical—we can’t pin these trends solely on AVs.

We must interpret these early findings with caution. The impacts we see are correlations, but not necessarily pure causation. AV companies didn’t launch in random locations—they targeted cities with favorable regulations and specific market conditions (e.g., tech-friendly laws or driver shortages). In other words, Phoenix, SF, LA, and Austin were primed for AVs, so any wage changes there partly reflect those unique local factors, not just robots hitting the road.

For example, Phoenix and Austin have relatively smaller legacy driver pools, so adding even a modest AV fleet might tip the supply-demand balance more than it would in New York City. California’s unique regulatory environment (permitting AV taxis in SF/LA) coincided with other local dynamics (tech industry recovery, tourism booms, etc.) that also affect ride demand and pay.

So while driver complaints that “robotaxis are cutting into our pay” are backed by data, we should interpret the “AV effect” as a localized, gradual shift rather than a simple cause-and-effect. The impacts vary by city—e.g., Austin and Phoenix fared a bit better than LA—in part because Waymo’s presence varies (only ~100 vehicles in Austin vs. several hundred in SF/LA) and each market’s competitive landscape is different.

The key takeaway is that AV deployments can coincide with downward pressure on driver wages and utilization, but the magnitude depends on local market conditions. So if we want to forecast the impact on the entire U.S. market, we have to be more cautious.

Long-Term Projections – Modeling AV Expansion and Gig Wages

What happens as autonomous fleets scale up further? Will gig driver earnings continue to erode in the long run, or even collapse entirely?

Academic research has started to tackle these questions with economic models—and the results are interesting. One important study by Auyon Siddiq and Terry Taylor (2022) modeled the ride-hailing market with a mix of human and autonomous drivers, exploring how a platform (like Uber) might optimally manage wages and AV deployment over time.

Their game-theoretic model delivered a nuanced picture of the future: as AVs become cost-competitive, platforms will gradually shift from human labor to AV fleets—but this transition can have counterintuitive effects on both wages and profits. Notably, Siddiq and Taylor found that introducing AVs can sometimes intensify competition and even reduce a ride-hailing firm’s profitability.

Why?

Because in a competitive market, if one company floods the system with driverless cars, it can lower fares to attract passengers (since AVs incur no per-trip wage). Competitors will likely match those price drops. But if prices drop too quickly, the wage savings from AVs may be outweighed by lost revenue. This means that robotaxis aren’t necessarily a guaranteed goldmine—aggressive expansion could trigger price wars that ultimately squeeze profits.

From the driver’s perspective, the model suggests an uneven path. During the early and middle phases of AV rollout, human drivers and AVs will co-exist (as they do now), and the platform must “balance” them—deciding how many rides to allocate to AVs versus humans, and what wage to offer to retain enough drivers. A surprising implication is that platforms might strategically raise driver wages while still reducing the number of human drivers active on the platform.

This counterintuitive result stems from how drivers anticipate AV expansion. When the platform raises wages, it is often because AVs are taking on a larger share of trips (for example, handling off-peak demand at low cost). Drivers then expect fewer ride opportunities overall, so some exit the market despite the higher pay. In effect, a higher per-trip wage can coincide with fewer drivers, since it reflects a shift toward an AV-dominated fleet.

A related study by Francisco Castro and Andrew Frazelle (2024) illustrates just how tricky the rollout can be from the platform’s side. They show that giving a platform the flexibility to deploy AVs alongside human drivers can actually backfire, because drivers anticipate how AVs will change their opportunities. In some cases, raising the per-trip wage reduces driver participation, since workers expect that more AVs on the road will leave them with fewer rides overall. Even more striking, platforms that expand their AV fleets too quickly may end up cannibalizing their own revenues, as cheap robotaxi trips substitute for higher-priced human-driven rides. The lesson from their analysis is that “getting out of your own way” requires restraint: scaling AV fleets gradually may preserve both profitability and driver stability better than an all-out push.

Such dynamics illustrate that driver wages in an AV-enabled market won’t just be a simple supply-and-demand outcome; they’ll also be shaped by strategic moves and expectations. Note that these studies hinge on the fact that drivers are somewhat strategic in making their choices to enter and exit the market, forcing the firms to react by increasing wages in these markets. The next few years will validate empirically how strategic they are.

Eventually, if AV technology becomes more affordable and widely accepted, the model predicts human drivers could be “priced out” entirely—the platform would employ an all-AV fleet once it’s more cost-effective than even minimum driver wages. At that point gig drivers would become obsolete (i.e., zero earnings), but in the interim decades, we’re likely to see them ebb and flow: platforms tweaking wages to ensure enough human coverage (especially for trips or areas AVs can’t handle yet), then reducing pay or hours as AVs expand.

Other projections also echo this “bumpy road” scenario. One analysis envisioned a mature U.S. robotaxi market where fares drop by 80% (because AVs slash operating costs), causing a massive expansion of rides but also a near-total displacement of human drivers. Uber’s own estimates, for instance, have hinted at a future where they operate millions of autonomous vehicles and fares per mile plummet—which would likely eliminate most driving gigs while creating some new fleet service jobs.

However, it’s not all doom and gloom.

These models also suggest some mitigating factors. If AV technology remains expensive or if regulatory barriers persist, platforms will rely on drivers longer and must pay enough to attract them. Competition for drivers (as opposed to riders) could still matter—for example, if two companies operate in a city, each might boost driver pay or incentives to ensure a sufficient workforce until AVs fully scale. And if robotaxis make rides so cheap that demand explodes, there could be more total trips such that drivers who remain (for niche services or premium rides) still find work (albeit likely at lower pay per trip).

Gig Workers and Generative AI: A Parallel Wage Squeeze

In 2025, no article can be written without asking what the impact of Gen AI on this industry is.

Believe me, I tried. E-mail servers send them directly to Spam.

But in the case of the gig economy, it’s a real question.

The wage compression now visible in robotaxi markets has a close parallel in digital freelancing, where generative AI is altering the economics of gig work. Recent academic studies offer some of the clearest early evidence. Personally, I seldom use Fiverr or Upwork, after being a frequent customer for many years. But in the same way correlation is not causation, an anecdote is not a data point.

As usual, I turn to academic studies, which use data to provide some of the clearest early evidence. A study by Hui, Reshef, and Zhou (2023) tracked millions of contracts on a leading online freelancing marketplace to measure the impact of ChatGPT’s release. They found that jobs most exposed to generative AI—such as writing, translation, and simple graphic design—saw measurable declines in demand within months. By contrast, categories less exposed to AI, like software engineering or accounting, were largely unaffected. The upshot: routine creative gigs contracted, while more technical or AI-complementary categories remained steady or even grew.

A related paper, “‘Generate’ the Future of Work through AI: Empirical Evidence from Online Labor Markets” analyzed project-level data from Upwork and found that text-related categories (copywriting, content creation) saw reduced postings after ChatGPT, while programming-related submarkets experienced smaller declines. High-rated freelancers lost some of their premium: the “flattening” of earnings distributions suggested that AI tools enabled lower-rated workers to compete more effectively, eroding the advantage of established stars.

At the same time, in their recent paper “Strategic Responses to Technological Change: Evidence from Online Labor Markets,” Yiu, Seamans, Raj, and Liu show how freelancers adapted in real time. Many specialized more narrowly, changed the types of contracts they pursued, or repositioned toward AI-complementary skills (e.g., chatbot development, AI content integration).

Surveys add another dimension. A 2024 global poll of over 4,000 freelancers (Freelancer.com 2024) found that 73% were already using generative AI in their work. The most common applications were AI-assisted writing (34%), AI-based design (19%), and data automation (16%). Over half reported higher earnings since adopting AI, largely due to productivity gains and the ability to take on more projects. Yet nearly as many expressed anxiety about long-term displacement. This duality reflects a “productivity dividend” for some and a “commoditization trap” for others: those who integrate AI raise output and sometimes income, while those sticking to traditional methods face declining demand and wages.

In fact, around 44% of freelancers today say they earn more than they did in traditional jobs, often because they have specialized expertise that commands premium rates. As AI automates the easy grunt work, human creativity, complex problem-solving, and interpersonal skills may become even more valuable. For example, while routine copywriting gigs fell, the same study saw rising demand for AI-related skills: jobs for machine learning experts and chatbot developers jumped as companies sought talent to integrate AI.

Gig platforms like Fiverr are responding by providing freelancers with AI tools to boost their productivity (e.g., an AI that helps a designer create drafts faster) rather than replace them. If that philosophy holds, some gig workers could actually benefit from AI by handling more gigs in less time or offering new services that merge AI and human insight. In an ideal scenario, AI could raise gig workers’ productivity, and if they have bargaining power or scarce skills, they could capture some of that gain as higher income.

The parallels to robotaxi cities are clear. Just as San Francisco drivers compensated for lower hourly wages by working longer shifts, many freelancers now work more hours or diversify skills to maintain income in the face of shrinking per-task pay. And just as rideshare incentives collapsed in AV hubs, freelance platforms are lowering effective pay for commodity tasks as AI fills the gap. Both settings show that automation does not instantly erase human gig work, but it compresses wages, shifts opportunity structures, and forces workers into longer hours or new niches.

These findings illustrate a critical point: automation impacts gig workers not only through substitution but through intensified competition and changing platform dynamics. In freelancing, AI blurred the skill hierarchy; in rideshare, AVs are reshaping utilization and incentives. In both cases, the outcome is wage pressure.

Final Thoughts

The gig economy isn’t just Uber and Lyft. It spans food delivery, freelance work, content creation, data labeling, and more. Reid Hoffman believes that by the mid-2030s the traditional 9-to-5 job will be largely replaced by gig-style, contract-based work—essentially, creating gigs out of “most jobs.” In Hoffman’s forecast, by 2034 we’ll trade single long-term jobs for portfolios of short-term roles, enjoying more flexibility but at the cost of stability.

And this is the problem.

The gig economy of the future—supercharged by AI and automation—is likely to be a double-edged sword for workers.

On one hand, it offers unprecedented flexibility and potentially global reach for talent.

On the other, it threatens to create a world of work where wages are fragmented, individualized, and relentlessly driven toward the lowest acceptable level by algorithms and competition.

The early impact of AVs on rideshare driver pay is a warning sign: even a relatively small penetration of automation can nudge earnings downward for those who rely on that work.

The challenge for society will be to harness the efficiency and innovation of AI and AVs without gutting the livelihood of millions. Achieving this means rethinking labor protections, supporting reskilling, but also ensuring people have enough information to make informed and strategic decisions, rather than be lured into the next gig or ride without realizing they’re on a downward spiral.