Chainlink Wants to Solve the Problem of Hallucinatory Pseudo-Results from AI Applications

Chainlink is implementing a new strategy to solve a major problem in artificial intelligence: hallucinatory AI systems. When large language models misinterpret data or generate incorrect new data, the consequences can be costly, especially in finance. Instead of relying on a single AI model, Chainlink is now taking a multi-model approach, using AI systems from OpenAI, Google, and Anthropic.

Laurence MORONEY, a Chainlink consultant and former head of AI at Google, explained that using multiple AI models instead of just one reduces the error rate. Each AI model is asked separately to analyze the same financial data. The system stores verified data on the blockchain, making it transparent, immutable, and secure. This consensus-based method prevents financial data from being corrupted by misinformation and increases the reliability of AI-generated data.

Chainlink’s approach aims to change this by reducing manual data verification and increasing financial accuracy. In a recent collaboration with leading financial institutions including UBS, Franklin Templeton, Wellington Management, Vontobel, and Sygnum Bank, Chainlink tested this AI-powered blockchain system. The results were promising, demonstrating a reduction in errors and inefficiencies in financial data.

Image: freepik

Designed by Freepik