Are Gig Platforms Biased Against Minority Freelancers?

Many freelancers who have experienced unfair treatment on freelancing websites say it can begin before a client ever sends them a message.

Many major platforms display each worker’s name and profile photo to prospective clients. The idea is to make freelancers seem friendly and trustworthy, but these features also introduce a window for unconscious bias.

Megan Williams, founder of the educational platform BlackFreelance.com, has argued that “Black-sounding” names and profile photos can influence hiring decisions.

“It’s pretty commonly known among Black freelancers that if you are visibly identifiable as Black and you use a photo, your contacts drop,” wrote Williams. She went on to say that she replaced her profile photo with an image of the Golden Gate Bridge and noticed a significant increase in responses and interview invitations.

What the research says about bias on freelancing platforms

A study published in Nature analyzed over 100,000 jobs and nearly 70,000 customer ratings on a home-services gig platform. The researchers discovered that under a standard five-star rating system, workers of color received an average of 4.72 stars, while White workers received 4.79 stars.

A 0.07-star gap might seem small, but the platform’s search and recommendation systems weighted workers’ ratings very heavily. That fractional dip translated to non-White workers earning just 91 cents for every dollar earned by their White counterparts.

As one article on the findings explained, “If a gig worker earns $40,000 a year, a 9% pay gap means they are taking home $3,600 less annually.”

Ratings and pay gaps can be exacerbated by “the algorithm”—in other words, by automated systems that recommend freelancers to clients.

This was backed up by another recent study, in which researchers found that automated systems on gig apps increased the ratings gap for non-White workers by about 80% and increased the earnings gap by 28%.

This may be because a human reviewer can spot a negative review left by an unfair client. An algorithm, on the other hand, may use that review as a reason to lower the worker’s profile visibility, which makes it harder to get good ratings to counterbalance the bad one.

For workers who receive unfair reviews due to prejudice, this creates a vicious cycle.

The upshot is that “minority workers have to expend more effort to get the same number of jobs as their White counterparts,” said Tristan Botelho, one of the lead researchers on the study.

How the algorithm changed Fiverr

Horror screenplay writer and narrative designer Nicky Jones, creator of Horror Movies and Stuff, says the landscape looked very different when he joined Fiverr—one of the most popular freelancing websites in the world—in 2009.

“Back then, the platform provided freelancers an opportunity to, quote, ‘be their own engine,’” Jones said. “What you put into your gig in terms of effort determined how lucrative it would become.”

While success was never guaranteed, Jones believes freelancers could build momentum if they delivered quality work. “When I first considered entering the world of freelancing, I accepted that bias was already baked into this tiny slice of the career cake,” he told us. “But in my opinion, you had a fair shot.”

Over time, however, Fiverr began replacing its staff with automated algorithms that weren’t as adept at understanding the nuances of client-freelancer relationships.

“It created very low morale among many of its freelancers,” Jones said.

One of his biggest issues is with how the algorithm treats every negative interaction as objective data, which ignores the existence of dishonest or unreasonable buyers who leave unfairly bad reviews.

“A method where one bad interaction can hurt my overall ranking, which in turn impacts both my traffic and income, puts me and many other freelancers at a serious disadvantage,” Jones said.

A different perspective

Freelancers from minority backgrounds, of course, aren’t a monolith, and not everyone agrees that the websites and apps they use encourage bias.

Gabrielle Seunagal, the founder of Gabrielle Writes and a freelance writer who has earned more than $300,000 on Upwork while maintaining a consistent five-star rating across more than 200 projects, has a somewhat different take on the ever-growing presence of the algorithm.

She told us she views it as a source of early-career friction rather than a permanent handicap.

“I certainly haven’t experienced any bias on Upwork,” Seunagal told us. “In my experience, platform algorithms—and clients themselves—favor freelancers with in-demand skills and vetted praise … my Upwork journey only got easier and smoother with more time and completed projects.”

For Seunagal, the algorithm isn’t inherently discriminatory, but it is highly competitive. It demands a grueling initial climb. Early on, she had to submit proposal after proposal. Today, clients come to her.

She believes that “frustrations with Upwork tend to smooth themselves out as you work, build on the platform, and prove yourself.”

Regardless of whether or not they’re inherently discriminatory, the algorithms on freelance and gig apps are controversial.

According to an investigation by the nonprofit group Human Rights Watch, these apps practice “dynamic wage-setting.” Because they aren’t bound by the federal minimum wage, they can track how long a worker is active on the app and how urgently they need income to calculate the absolute lowest payout they are likely to accept for a given job.1

The system often leads to different workers getting paid different amounts for the exact same labor. This means that those with shakier finances and fewer alternatives—who are more likely to be from minority backgrounds—operate at a disadvantage.

In an article for Inequality.org, writer Michael Huggins referred to this as surveillance pricing that forces these workers to accept a desperation wage.

What’s more, unlike white-collar freelancers on Upwork or Fiverr, app-based workers (like Uber drivers) have no room to negotiate for higher rates. They must either accept the algorithmically generated rate or decline the work entirely.

Legal disputes over protections for gig workers

Some legal scholars, such as Veena Dubal, argue that this practice builds on a history of excluding certain workers from traditional protections while pretending otherwise.

During the nationwide racial justice protests of 2020, gig platforms heavily co-opted social justice language to claim that their platforms empowered minority workers. For instance, Uber called themselves anti-racist and Lyft ran ads featuring Maya Angelou’s poetry.

However, simultaneously, these companies—among others—spent over $200 million backing Proposition 22 in California, which was a measure that exempted app-based companies from providing standard employee protections.2

In one of Dubal’s legal analyses, she argued that Prop 22 operates exactly like the discriminatory “wage codes” of the 1930s. Back then, lawmakers intentionally excluded majority-minority professions—like agricultural and domestic workers—from new minimum wage laws, allowing employers to legally pay Black workers 20% to 40% less than workers in white industries.

Dubal argued that Prop 22 is a digitized form of this “legalized racial subordination.”

By categorizing gig workers as a separate, “second-class” tier of independent contractors, she wrote, the law allows platforms to pay workers for only some of the hours they actually work (ignoring the time they spend waiting for gigs).

Moreover, because Prop 22 stripped away the requirement for a guaranteed hourly rate, it gave dynamic wage-setting algorithms the legal loophole they needed to run their “surveillance pricing.”

In Dubal’s view, this allows the machine to undercut a disadvantaged workforce, which—particularly in Californian cities like San Francisco—is composed primarily of immigrants and people of color.

In the face of algorithmic oversight, successful creatives and other freelancers find their own ways to protect their livelihoods.

For Seunagal, that means staying ahead of the curve. As AI disrupts the writing industry, she’s focused on finding ways to leverage it to justify her premium rates, such as producing AEO (Answer Engine Optimization) content to help clients rank in Google AI Overviews.

She also ensures she isn’t entirely dependent on Upwork, regularly sourcing clients through her own website and LinkedIn.

Jones shares a similar philosophy. He diversifies his income through his horror media brand. He pointed out that pulling clients off-platform helps avoid the “fee drain” that comes from marketplace commissions.

Protecting yourself from unfair reviews

On freelance platforms like Upwork, to protect your algorithmic standing from problem clients, veteran freelancers suggest a few practical safeguards:

• Establish explicit scopes of work: Define precise deliverables, milestone deadlines, and revision limits in writing before starting. Clear, objective expectations leave less room for subjective, bias-driven downgrading at the end of a project.
• Check in early and often: Request feedback mid-project to address communication gaps before they escalate. This forces clients to articulate specific issues early on, rather than quietly leaving a lower star rating after the contract is closed.
• Encourage text-based reviews: Written testimonials carry more weight with future clients than simple star ratings. Actively ask satisfied clients to write detailed feedback, which helps offset minor, automated fluctuations in your numerical score.

Jones offered a final piece of advice. “Never work with doubters,” he said. Avoid clients who begin negotiations by questioning your rates or experience. “They will almost always weaponize their skepticism into forcing you to work below rate, or worse, for free.”

Can gig workers follow these strategies?

Unfortunately, these safeguards are largely limited to creative freelancers. For the millions of gig workers operating on delivery or ridesharing apps, negotiating rates, establishing contracts, or taking clients off-platform is essentially impossible.

These workers have no room to negotiate and must either accept the app’s algorithmically generated rate or decline the work entirely.

Despite the ongoing research into algorithmic bias, there is little consensus on a fix.

Seunagal remains pragmatic about the competitive nature of the space: “I’m not sure equitable outcomes for everyone can be guaranteed.”

But Jones argues that platforms owe their workers a more level playing field, perhaps by re-introducing human oversight for client disputes. “Not every dispute between freelancer and client is the same,” he noted, emphasizing that machines simply cannot contextualize unfair situations.

What the research says about potential solutions

Interestingly, researchers have already proven that simple platform tweaks can yield massive results. In the Nature study, when the platform abandoned the traditional five-star scale in favor of a simple thumbs-up or thumbs-down model, the racial disparities in ratings (and the resulting earnings gap) virtually disappeared.

It’s not a perfect fix. A biased client could in theory still leave a thumbs-down after receiving stellar service. However, the study found that this prevented more subtle biases from creeping into a rating that an algorithm could magnify into a systemic pay gap.

Uber Eats, for example, has implemented this system, while also offering quick-reply chips where customers can give additional positive or negative feedback.

Uber Eats’s adoption of this approach suggests that it’s feasible, at least for ridesharing and delivery apps.

At the same time, simplifying customer ratings may only fix one part of the issue.

While frustrations with algorithmic oversight aren’t universal (as Seunagal pointed out), research and freelancers like Jones indicate they are a common reality.

If gig companies want to deliver on the promise of a meritocracy, they’ll need to tackle the underlying biases in their recommendation algorithms, dynamic wage-setting models, and overarching labor policies.

If they don’t, all the signs indicate that for a significant percentage of minority freelancers and gig workers, algorithmic bias will continue to be a sticking point for years to come.