AI model training needs are changing, with subject experts pushing aside generalist data labelers (www.businessinsider.com)

A recent report by HireArt highlights a significant shift in AI model training needs, as the increasing complexity of AI systems necessitates subject-matter experts over generalist data labelers. The 2025 AI Trainer Compensation Report reveals that the demand for nuanced reasoning and specialized knowledge in domains such as law, engineering, and medicine has transformed data labeling from a basic task into a critical cognitive function. Consequently, highly skilled trainers are now essential for enhancing model intelligence, safety, and trustworthiness, thereby reflecting the maturation of the AI industry. The report indicates a stark contrast in compensation: while entry-level trainers in the U.S. earn between $12.50 and $15.50 per hour, subject experts can command rates exceeding $100 per hour, with medical professionals earning between $60 and over $180 per hour, and experts in engineering and law making upwards of $150. This evolution underscores the growing recognition of the value that specialized knowledge brings to AI development, marking a shift towards a more professionalized and skilled workforce in the AI/ML community.