How human expertise is quietly powering AI

AI development increasingly relies on expert human trainers to improve performance in complex fields. Companies specializing in AI training enlist freelance experts from diverse backgrounds to create training data and evaluate AI responses. Scale AI and Invisible Technologies are key players, building platforms to help AI learn various tasks. These firms offer access to vetted experts and promise quick data production in niche areas. Training tasks include providing prompts, answers, rubrics, and corrections, with a growing need for highly specialized experts, often with advanced degrees. Experts are paid competitively, with some earning hundreds of dollars per hour, motivating participation alongside intellectual curiosity. Professionals like physicians and professors contribute by identifying AI failures and guiding ethical development. Companies like Mercor and Micro1 focus on sourcing and evaluating talent, aiming to match experts with enjoyable tasks. Surge AI emphasizes the importance of human guidance in aligning AI as standards for accuracy increase. While some AI training contractors have faced gig economy issues, the field is poised for growth with continued AI investment. This expansion includes enterprise work and training AI for physical world interactions.