AI Model Beats GPT

Bridgewater and Thinking Machines Lab have developed an open-source AI model that outperforms leading commercial models in analyzing financial documents, achieving nearly 85 percent accuracy at a fraction of the cost. This breakthrough demonstrates the power of fine-tuning AI models with internal expert knowledge, allowing companies to develop powerful AI solutions without sharing sensitive information. The model’s success is significant because it automates the constant stream of small, repeated judgment calls that investors face daily. Why it matters: this development shows that companies can leverage their own data to create effective AI solutions, reducing reliance on large providers and sensitive information sharing, which is a game-changer in the AI landscape.