OpenAI Expands ChatGPT Training Across 400+ Jobs to Build Expert-Level AI
OpenAI is expanding ChatGPT beyond coding and writing into hundreds of specialized professions. According to a recent investigation, the company has enlisted thousands of freelance professionals to train the system across more than 400 occupations. These roles range from commercial pilots and pharmacists to soil scientists and agricultural managers.
As part of this initiative, the program known as Project Stagecraft plays a central role. Moreover, Handshake AI manages the effort while shifting from its earlier career platform model. Under this structure, between 3,000 and 4,000 freelancers contribute their expertise. In addition, they earn at least $50 per hour to build detailed personas and create realistic task prompts based on their daily work.
Meanwhile, some expert roles command rates as high as $500 per hour, although not all connect directly to OpenAI projects. A training manual further instructs contributors to prioritize knowledge work instead of manual labor. Therefore, the initiative focuses on mapping economically valuable tasks and evaluating how effectively ChatGPT handles domain-specific queries.
Industry Shift Toward Expert-Level AI
Project Stagecraft highlights a broader shift across the AI industry. As general-purpose capabilities mature, companies now focus on capturing deep professional expertise. Consequently, they aim to bridge the gap between generic responses and expert-level decision-making.
This transition has also accelerated the growth of third-party data firms such as Scale AI, Surge AI, and Mercor. These companies actively recruit skilled professionals to generate high-quality training data. As a result, AI systems gain more nuanced and specialized knowledge across industries.
However, this strategy also raises concerns among workers. One freelancer working in a niche field told Business Insider, “We all were aware that we were basically training AI to replace us”. Thus, the remark reflects broader anxieties across sectors such as agriculture and nursing.
Data Practices and Ongoing Concerns
At the same time, the initiative follows earlier reports about similar data collection practices. For example, a January report by Wired revealed that contractors were asked to upload real work materials. These included documents, code, and presentations from both current and past roles.
As a result, companies could establish performance benchmarks for next-generation AI models. However, neither OpenAI nor Handshake AI responded to requests for comment on Project Stagecraft. Therefore, questions remain about data sensitivity and transparency as the program continues to evolve.
