Artificial Intelligence in Banks is a secret project for OpenAI to enter the world of financial analysis..
Bloomberg has revealed a secret project within OpenAI, the developer of ChatGPT, codenamed Mercury or Mercury, that aims to train its AI models to perform traditional tasks carried out by novice financial analysts in investment banks.
According to the report, the company has hired more than 100 former bankers from major institutions such as JPMorgan, Morgan Stanley and Goldman Sachs to train its models in building financial models, preparing IPOs and restructuring, in accordance with the strictest standards of coordination and audit adopted in the financial sector.Participants in the project are paid $150 per hour for financial models and prompts, which are used to train AI to simulate the exact workings of investment analysts, according to the Bloomberg report, and are required to strictly adhere to banks' coordination rules, including the size of margins and the way percentages are written in italics.
The project aims to enable AI systems to perform financial analysis and presentation tasks in seconds, without the need for continuous human supervision, which is expected to reduce the cumbersome workloads typically shouldered by new analysts in investment banks, some of whom work more than 100 hours per week.On the other hand, the project raises questions about the future of jobs in the financial sector, as the success of the models in performing the tasks of analysis and coordination accurately could mean dispensing with an entire class of junior analysts who form the base of the professional pyramid in investment banks.The sector has traditionally relied on a professional grading system that starts with young analysts and progresses to executive partner positions through years of field experience.Thismove by OpenAI comes at a time when the banking sector is witnessing a massive digital revolution in the adoption of artificial intelligence.