High-end AI training market surges as professionals sell expertise

A new gig economy is emerging as experts like lawyers and doctors earn up to $400 per hour to train AI models. This shift toward high-level data training, exemplified by the $20 billion platform Mecko, moves beyond simple labeling to the replication of professional judgment and logical processes.

As artificial intelligence (AI) technology advances, the landscape of the labor market is changing rapidly. While AI training in the past was limited to simple repetitive tasks (labeling), such as distinguishing objects in photos, it has now entered a stage of purchasing the judgment and know-how of highly skilled professionals like doctors and lawyers. A new "gig economy" is forming where experts receive high compensation, reaching hundreds of dollars per hour, in exchange for training AI with their knowledge.

"My Know-how to AI"... The Emergence of a High-Priced Data Market Targeting Professionals

According to a video from the YouTube channel "Understanding: All the Knowledge in the World," professionals such as doctors and lawyers are participating in large numbers in high-paying, short-term part-time jobs to train AI models in global AI data labor markets, including the United States. Cases reported by foreign media such as Bloomberg show that civil rights lawyers perform coaching duties—asking AI complex legal questions and correcting errors in its answers—earning up to $400 (approximately 530,000 won) per hour. They perform this work by explaining why an AI's answer is wrong, adding insufficient explanations, or sharing their own experience and know-how.

These high wages go beyond simply delivering information. AI companies do not want to be told just the correct answer; they want to teach the "judgment process" and "logical basis" an expert uses to reach that answer. For example, the targets are areas requiring high expertise, such as radiologists who find minute calcifications in chest X-rays or engineers who catch complex coding errors. From a corporate perspective, a model trained on an expert's judgment criteria can hold a powerful competitive advantage in the market. In fact, recruiters are desperate to hire "real players," such as radiologists to point out minute abnormalities in chest X-rays or engineers to solve high-level coding errors, through platforms like LinkedIn.

The case of Mecko, a data training platform for professionals, clearly demonstrates this shift. Mecko has grown rapidly, reaching a valuation of approximately $20 billion (approximately 27,000,000,000,000 won) within just three years of its establishment. Various professional demands exist on this platform, such as recruitment for studies targeting parents for research at Stanford University ($110 per hour) or toxic chemistry experts ($65–$75 per case). In particular, online surveys targeting urologists and oncologists to ask for professional opinions on selecting and evaluating prostate cancer treatments offer compensation of $115 (approximately 150,000 won) per case, requiring strict qualifications such as clinical experience within the last year and a doctoral degree.

The End of Simple Labor and the Paradox of "Knowledge Replication"

The craze for high-paying part-time jobs among professionals is deepening labor market polarization while simultaneously creating a paradoxical situation that shakes existing labor structures. Platforms like Mechanical Turk, which was operated by Amazon in the past, provided jobs for low-wage blue-collar workers, such as image labeling (e.g., drawing a square around a bicycle in a photo) or voice transcription. However, the recent termination of operations by such platforms signifies that AI has moved beyond simple data distinction. This is because AI is now capable of performing much higher-level tasks, such as reviewing legal documents or detecting cancer cells, beyond simple object recognition.

The problem is that the more professionals are paid to train AI, the greater the paradox that AI is more likely to replace those very professionals' jobs. While past labor was a concept of "renting" human time and labor, current AI training is closer to an attempt to "download" and permanently possess an expert's knowledge and judgment capabilities into a model. If an expert provides their know-how to an AI, companies can perform similar levels of work without that expert through an AI model that has replicated that knowledge. The video explains this as "enabling the continuous use of an expert's knowledge and judgment capabilities by training them into AI."

In this process, while the difficulty of labor is increasing, a phenomenon where jobs are decreasing is appearing. Simple data workers are being required to perform higher-level tasks that AI cannot handle itself (such as fact-checking or comparing chatbot answers), and among professionals, the anxiety of "Am I not training my own replacement?" is approaching as a real threat. Because the value of the data provided by experts lies not just in knowing a lot of information, but in providing the criteria for judgment—such as "what must be verified to avoid errors"—the scale of compensation is formed at a level entirely different from simple labor.

Areas AI is Penetrating: Issues of Values and Faith

The influence of AI is extending beyond the professional work areas of experts into the realms of human values and faith. While people previously consulted clergy for life concerns or religious questions, an increasing number of people are now demanding logical and systematic answers from AI. This is why analyses suggest that the rational and practical advice provided by AI may partially replace the authority or roles of existing religious leaders.

However, the limitations of AI's answers are also being pointed out. AI can be biased toward specific values depending on its trained data and design methods, and sometimes shows a tendency to provide answers that users want to hear. This leaves room for conflict with the moral judgments or religious teachings emphasized by religion. Accordingly, a new phase is unfolding where AI and human values clash and coexist, as seen in movements within religious circles to develop their own AI chatbots to reflect doctrine and religious values.

Source: original video (YouTube)