Mercury" (Mercury) is an ambitious internal project aimed at automating repetitive tasks typically handled by junior investment bankers, targeting the core operations of Wall Street.
According to Bloomberg, OpenAI has secretly hired over 100 former bankers and financial experts, using their professional financial modeling experience to train its artificial intelligence system. The ultimate goal of the project is to teach AI how to independently generate complex financial models, potentially eliminating the tedious basic work that junior analysts spend hours on.
Elite Team and Unique Work Model
The team members of the "Mercury" project are a gathering of elites, including former employees from major financial institutions such as Goldman Sachs, JPMorgan Chase, Morgan Stanley, Brookfield, Evercore, and KKR, as well as MBA students from Harvard University and MIT.
These experts were recruited through third-party suppliers, with flexible working hours and an hourly rate of approximately 150 dollars. Their core task is to build a simulated merger, restructuring, or IPO transaction financial model each week. They convert the AI-generated results into Microsoft Excel by writing simple prompts, and continuously improve the model based on feedback, all operations following strict industry standard formats.
AI-Driven Recruitment and Training Process
The recruitment process for this project is almost entirely automated. According to sources, candidates first undergo a 20-minute interview conducted by an AI chatbot, followed by knowledge and modeling tests. The models they submit are reviewed, and the feedback from the reviewers is directly entered into OpenAI's training data. The knowledge gained from these practical examples aims to empower the AI system to independently generate high-quality financial models.
An OpenAI spokesperson confirmed to Bloomberg that the company does collaborate with experts from various fields to improve and evaluate the capabilities of its models, and noted that these experts were recruited and paid by external suppliers.
The exposure of the "Mercury" project signals that conversational AI is rapidly moving from general areas into highly specialized financial services, posing significant challenges to the future work models of junior investment banking positions.
