AI Side Hustles Misleading — the headline number is real but unrepresentativ

By the Income Reality Check Editorial Team · Updated 2026-11-12 · 7 min read

AI training gigs pay $40–$100/hr? What Mercor, Outlier and DataAnnotation actually hand over

Verdict: Misleading — the headline number is real but unrepresentative. The sites are legitimate and do pay cash, but the four-figures-a-week framing belongs to a small slice of scarce specialists, not the no-experience beginner the video is speaking to.

In “How To Make Money THIS WEEK,” Beverly Dines — a self-described chief operating officer turned “chief empowerment officer” — tells 203,000-plus viewers they can get paid “$40 an hour, $50 an hour, $100 an hour with no technical experience” by training AI on five vetted sites, and can apply “within the next 24 hours.” The platforms she names are real. The pay numbers exist. What’s missing is who those numbers are for, and how steady the work underneath them actually is. Short answer: the gig is genuine; the income picture she paints is the top of a very lopsided curve.

What the video actually claims

Dines frames this as a “gate-kept side hustle” that doubles as a career pivot. The pitch has three parts. First, get trained for free — she points to Mercor’s learning center, “YouTube University,” and budget one-on-one tutoring on Preply (she suggests “15 bucks an hour or 20 bucks an hour”) rather than a four-figure bootcamp. Second, “get paid to build your resume” by signing up for five sites she says she vetted herself: Mercor, Outlier, DataAnnotation, Handshake, and Meridial (powered by Invisible Technologies). Third, use that logged experience to land a part-time or full-time “AI analyst” or “AI prompting” role in your existing industry.

To her credit, Dines is careful where a lot of creators aren’t. She repeatedly tells viewers not to pay for expensive courses. She says the work is 1099 contractor work and that a legitimate site will have you fill out a W-9 — a tax detail most hype videos skip entirely. She even warns that stacking certifications can turn a “side hustle” into “a side job to get a side job.”

So this isn’t a scam pitch. It’s an optimistic one. And optimism is where the reporting has to start.

What the method actually requires

Here’s the first gap. Dines says “sign up for all of them” because onboarding times vary — which is true, but it quietly concedes the core problem with this work: you don’t control when (or whether) tasks appear. Independent reviews of these platforms in 2026 describe the same pattern over and over. Projects ramp up to 40 hours of available work for two weeks, then abruptly pause when a model checkpoint finishes training. Queues go empty for weeks. Accounts get deactivated without notice. One contributor reported completing more than 80 onboarding processes, tutorials, and assessment tasks — all unpaid — before seeing real work.

That unpaid-onboarding detail matters, because it’s exactly the kind of cost the U.S. Federal Trade Commission has gone after. In August 2025 the FTC sent more than $6.7 million to 98,254 consumers harmed by gig-work firm Arise Virtual Solutions, which advertised remote customer-service work “up to $18/hour.” The agency found most agents spent time and money on mandatory equipment and training, then earned less than the promised rate. The lesson isn’t that Mercor or Outlier are Arise — it’s that “up to $X an hour” is a ceiling, and the real number lives after the unpaid hours.

The pay reality, platform by platform, looks roughly like this in 2026:

Then there’s tax. Because these are 1099 roles, you owe self-employment tax of 15.3% on net earnings, and the IRS requires you to file once net self-employment income hits $400 — even for a side gig, even if no 1099 arrives. You may also owe quarterly estimated taxes. A “$50/hour” gig is not $50 in your pocket.

Who actually wins this game?

Follow the money and a clear profile appears. On Mercor, the eye-popping rates attach to narrow expertise: the platform has listed oncologists at $350–$450/hour and other clinicians at $120–$150/hour. Those aren’t “no experience” jobs — they’re scarcity-priced because AI labs need a credentialed doctor, lawyer, or senior engineer to grade a model’s output, and there aren’t many of them willing to do piecework.

The people who clear meaningful money here tend to share two traits: an unusually specific domain skill the labs will pay a premium for, and enough financial cushion to ride out the dead weeks between contracts. A generalist rating chatbot responses for $16 an hour, with no guarantee of hours next month, is playing a very different game than a radiologist annotating scans. Dines actually gestures at this — her whole “domain expert” framing is correct — but she pairs it with “$100 an hour, no technical experience,” and those two ideas can’t both be true for the same person.

What you’d realistically earn

The advertised average across AI-training gig work in 2026 lands around $31/hour, but after unpaid screening, empty queues, and rejected tasks, generalists bank closer to $22 — and median annual data-worker earnings sit under $23,000. That tracks with the broader government numbers: the U.S. Bureau of Labor Statistics reported that in 2023 there were about 11.9 million independent contractors, and contingent workers earned a median of $838 a week versus $1,137 for workers in traditional jobs. Contract work pays less and less reliably — that’s the structural reality, not a knock on any one site.

For a realistic beginner, expect something like $0 for the first week or two of unpaid onboarding, then sporadic blocks of $15–$30/hour work that may dry up without warning. Could a motivated person net a few hundred dollars in a good month, or more during a busy project? Yes. Could they hit the “$100 an hour” line Dines opens with? Only with a credential the video assumes you don’t have. If your plan is steady, predictable income, this isn’t it — a point echoed by job-hopping readers who’ve chased online jobs that “hire anyone with a pulse” and found the hiring easy and the hours erratic.

One more thing you should factor in: how these companies treat the people doing the work. In October 2026, Scale AI — the company behind Outlier — agreed to a $12.5 million settlement of California claims that it misclassified taskers as contractors and cut their pay when work ran past assigned time limits. Scale denied wrongdoing. The suit alleged workers were recruited at $25–$40/hour, then docked. It’s a live legal fight, and it’s the clearest signal yet that the gap between advertised and delivered pay on these platforms is real enough to litigate.

Who this is (and isn’t) for

This makes sense if you have genuine domain depth — nursing, law, software, finance, a specialized trade — a few flexible hours a week, and the patience to treat it as top-up money rather than a paycheck. It also works as a low-stakes way to learn how these models behave, which is a real resume line in 2026. It does not make sense if you need a reliable amount by Friday, have no specialized background, or are counting on it to replace a salary. If that’s you, the career-change framing in the video’s third act — pivoting into an “AI-adjacent” role in your own industry — is actually the more durable play, and it rhymes with advice for anyone starting a career over from scratch.

What to remember

Dines is selling a real door, just a narrower one than the thumbnail suggests. Mercor, Outlier, DataAnnotation, Handshake and Meridial exist, pay cash, and issue tax forms — that part is honest, and so is her refusal to upsell a bootcamp. But “$40–$100 an hour, no experience” fuses the best-case specialist rate with the beginner’s entry point, and those two numbers belong to different people. Treat it as flexible, taxable, unpredictable gig income with a useful learning curve attached, and you’ll be calibrated correctly.