Chainlink has introduced a brand new strategy to tackling the difficulty of AI hallucinations when massive language fashions (LLM) generate incorrect or deceptive data.
Laurence Moroney, Chainlink Advisor and former AI Lead at Google, defined how Chainlink reduces AI errors through the use of a number of AI fashions as an alternative of counting on only one.
Chainlink wanted AI to research company actions and convert them right into a structured machine-readable format, JSON.
As a substitute of trusting a single AI mannequin’s response, they used a number of Giant Language Fashions (LLMs) and gave them totally different prompts to course of the identical data. For this, Chainlink makes use of AI fashions from suppliers like OpenAI, Google, and Anthropic
The AI fashions generated totally different responses, which had been then in contrast. If all or a lot of the fashions produced the identical outcome, it was thought-about extra dependable. With this course of, the chance of counting on a single, probably flawed AI-generated response decreases.
As soon as a consensus is reached, the validated data is recorded on the blockchain, guaranteeing transparency, safety, and immutability.
This strategy was efficiently examined in a collaborative challenge with UBS, Franklin Templeton, Wellington Administration, Vontobel, Sygnum Financial institution, and different monetary establishments, proving its potential to scale back errors in monetary information processing.
By combining AI with blockchain, Chainlink’s technique enhances the reliability of AI-generated data in finance, setting a precedent for bettering information accuracy in different industries as properly.
Additionally Learn: Aptos Adopts Chainlink Standard for Secure Data Feeds
