Atomic Canyon,
استارتآپ آمریکایی Atomic Canyon با جذب سرمایه اولیه ۷ میلیون دلاری و آغاز همکاری با نیروگاه هستهای Diablo Canyon، تلاش خود را برای متحولکردن صنعت انرژی هستهای با بهرهگیری از هوش مصنوعی آغاز کرده است. این شرکت نوپا تمرکز خود را بر توسعه یک سیستم جستوجوی هوشمند گذاشته که بتواند دسترسی سریع و دقیق به اسناد فنی را برای مهندسان، نیروهای تعمیراتی و ناظران انطباق تسهیل کند.
The founder of this startup, Terry Ladredale, who previously worked in the healthcare sector, realized the massive and unmanageable volume of documents in the nuclear industry after interacting with staff at a power plant near his home in California. For instance, the Diablo Canyon plant alone has over two billion pages of archived documents. This fact sparked the idea of using artificial intelligence algorithms based on the “Retrieval-Augmented Generation” (RAG) method in his mind.
At the beginning of the journey, the AI models faced significant challenges because the specialized vocabulary and terminology of the nuclear industry were not understandable to them. However, with the support of the Oak Ridge National Laboratory, this startup was able to receive 20,000 GPU hours for training and optimizing its model, overcoming this technical barrier.
The main goal of the Atomic Canyon team now is to combine artificial intelligence and human effort to produce accurate and referenceable drafts of nuclear documents; a process that can significantly reduce time, cost, and the likelihood of errors in this sensitive industry.
This initiative comes at a time when many countries are seeking to modernize their nuclear energy infrastructure, and the nuclear industry, especially in the U.S., faces challenges such as a shortage of skilled workforce and high operational costs. Using artificial intelligence in this field can help efficiently digitize old documents, reduce administrative burdens, and facilitate knowledge transfer between generations. Experts believe that RAG models, equipped with precise retrieval of relevant data at response time, can play a key role in sensitive industries like nuclear energy; as they offer both high accuracy and source traceability, which is crucial for compliance with international safety standards.