Interested in more careers-related content? Check out our new weekly Work Life newsletter. Sent every Monday afternoon.
Rachael Andrews started her career as an on-air personality for an Ottawa radio station and continued part-time voice work after transitioning into the banking industry a few years later.
So when she stumbled upon a LinkedIn ad for a voice acting contract early last year, she was intrigued.
“When you see really interesting job postings on LinkedIn, sometimes they’re almost too good to be true,” Ms. Andrews, 49, says.
The company was Mercor, an San Francisco-based company that pays experts to train AI models.
At first, Ms. Andrews spent hours recording scripts for the AI startup via videoconference.
“They [offered] as many hours as I could grab at that time, because they were just getting off the ground and there was a lot to do,” she recalls.
Today, Ms. Andrews works about 30 hours a week for Mercor – outside of her full-time job at the Tourism Industry Association of Canada in Ottawa – managing and coordinating a network of experts across various AI research projects worldwide.
The single mother of seven-year-old twins has also referred more than 100 others to the platform, including her kids’ pediatrician, former colleagues from the finance industry and her next-door neighbour.
“It changes lives, if you do it right,” she says.
The money earned from the AI side gig has helped her pay off a new car and take her kids on vacation.
The US$10-billion company, founded three years ago, is one of many AI training platforms offering cash in exchange for expertise on a flexible, temporary contract basis. Others offering similar work include Alignerr, Data Annotation, Outlier, Mindrift, Handshake, RemoExperts, RWS and Vancouver-based Telus Digital AI.
Mercor distributes more than US$3-million in payment to roughly 30,000 “experts” every day, according to the company. The hundreds of positions on its website include openings for management consultants (US$100 per hour), mechanical engineers (US$90 per hour) and French-Canadian voice actors (US$50 to $150 per hour).
“We’ve had people work with Mercor who are comedians, chess experts, gamers, sommeliers,” says Heidi Hagberg, the San Francisco-based company’s head of communications. “People go to ChatGPT and ask, ‘what’s the wine pairing recommendation with my steak?’ and the model doesn’t know that answer unless there’s somebody guiding and educating it.”
Mercor doesn’t publicly disclose its clients, but reportedly sells the data it gathers to many of the world’s top AI companies, including OpenAI, Anthropic and Meta.
“We don’t share names but it’s like every frontier [AI] lab and an increasing number of top-tier enterprises,” says Ms. Hagberg when asked to verify the clients listed in the report.
Despite the variety of job opportunities posted on its website, Ms. Hagberg says the company is primarily seeking medical, legal, software engineering, consulting and finance professionals. She says pay averages about US$120 per hour, with contracts ranging from a few days to more than six months.
Ms. Hagberg says those who share their expertise with the company are typically seeking to supplement their incomes with a flexible remote opportunity, increase their AI proficiency or join for more altruistic reasons.
For example, Ms. Hagberg says one Bay Area emergency room doctor told her that after decades of seeing patients who received inaccurate information online, she felt a responsibility to ensure the next wave of technology is more reliable. She also wanted to help make accurate medical information more accessible.
“She sees it as a higher purpose and a personal responsibility,” Ms. Hagberg says.
Professionals are typically asked to complete one of three tasks: reviewing AI outputs and ranking their accuracy and quality; outlining how they do their jobs in detailed steps; or using a rubric to grade how well the model executed a more complex process or series of tasks.
Ms. Hagberg describes this type of work as the next phase in the technology’s development.
“Ten years ago, [training AI] was drawing boxes around stop signs, and it was very redundant, very low-skilled work,” she says. “Now companies like Mercor are focused on highly educated, highly specialized work, and that is because AI models have evolved to become incredibly smart.”
To some, that may sound like paying humans to train their robot replacement, but Ms. Hagberg doesn’t see it that way.
“When productivity rises, so does the appetite for what people can do, what they can build,” she says. “Humans aren’t just going to sit back and be like, ‘okay, AI is going to do everything,’ because there’s just more problems to solve.”
With billions pouring into the technology, these types of roles also offer everyday professionals an opportunity to cash in on the AI gold rush, including Ms. Andrews.
“What I’ve said to people at my day job is, AI isn’t coming for your job, it’s coming for tasks that can be repetitive, manual and things that can be done faster,” she says. “I don’t know that AI will ever fully replace people; I think it will change how we work.”