The Urgent Need for Recruiting People To Train More AI
The race to build advanced AI models has revealed a critical bottleneck: the technology still needs significant human guidance. Recruiting people to train more AI has unexpectedly become the fastest-growing job market segment in Silicon Valley, moving from a niche service to a core driver of the AI boom. This shift is happening because models like GPT-5 and Gemini 2.5 Pro require high-quality, expert human feedback—known as data labeling or reinforcement learning with human feedback (RLHF)—to become truly intelligent and accurate. Without this human layer, the models struggle to grasp complex, nuanced, and ethical concepts.
Mercor’s Pivot to Recruiting People To Train More AI Pays Off
One company exemplifying this monumental trend is Mercor, a startup founded by three 22-year-olds. Initially building an AI recruiter, the company quickly discovered a more lucrative path: sourcing humans to train AI models. This rapid pivot was fueled by a dramatic market event: Meta’s acquisition of a significant stake in data labeling giant Scale AI. This move raised concerns among other major AI labs about Scale’s neutrality, creating an immediate, massive opening for a competitor focused purely on recruiting people to train more AI.
Expertise Drives the Demand for Recruiting People To Train More AI
Mercor’s success is rooted in its focus on what it calls the “Expert-as-a-Service” model. They aren’t simply hiring for basic data labeling tasks; they specialize in Recruiting People To Train More AI who are highly credentialed, such as PhDs, lawyers, management consultants, and financial experts. These domain experts are crucial for training the most advanced reasoning models, which need to be taught complex, multi-step problem-solving. This high-skill work commands premium compensation, with contractors earning between $90 and $200+ per hour.
A Multi-Billion-Dollar Industry Is Born
The financial indicators for this new sector are explosive, validating the shift towards actively recruiting people to train more AI. Mercor’s valuation soared from $250 million in late 2024 to a massive $10 billion after its latest funding round, positioning it as a key player. The company is now reportedly on track to hit $500 million in annual recurring revenue, growing at nearly 60% month-over-month in its fastest growth period. This success demonstrates that the market views human data annotation as a vital, high-margin piece of the multi-trillion-dollar knowledge work automation puzzle.
Addressing Bias While Recruiting People To Train More AI
A key challenge in the new AI economy is ensuring fairness and mitigating bias. Mercor contends that its AI-driven matching platform helps safeguard against traditional human recruiting biases by screening candidates based purely on skills and expertise, preventing the AI from seeing identifiers like name, gender, or race. This ethical consideration is critical, as the human feedback used in recruiting people to train more AI directly influences the model’s future behavior. Transparency and fairness are essential components for maintaining trust and complying with evolving global AI regulations.
The Future: Automation’s Human Precursor
While the long-term goal for companies like Mercor is to use AI for all job matching, the immediate future is undeniably in the human-in-the-loop economy. Recruiting people to train more AI will remain necessary as long as models need to master sophisticated concepts and respond to multi-faceted user requests. Investors and founders alike see this as a necessary, long-term phase of development, betting that human experts will be needed for years to come to ensure the next generation of AI is safe, effective, and ethically sound. The new job is teaching the machines that will eventually transform our own jobs.
Credit: Forbes.com
