Міждисциплінарний підхід до енергоживлення штучного інтелекту (ШІ)

• Дата: 2025
• Автор: Prazian, M.
• Формат: Стаття
• Мова: English
• Опубліковано: State Scientific and Technical Center for Nuclear and Radiation Safety 2025
• Онлайн доступ: https://nuclear-journal.com/index.php/journal/article/view/1215
• Назва журналу: Nuclear and Radiation Safety

Developing sustainable and resilient ICT-energy networks is crucial for powering artificial intelligence (AI). Bottlenecks in power infrastructure – such as generation, transmission, storage, and operational stability – can hinder the growth of data centers. Powering AI relies on various disciplines and factors, including technology, human elements, regulations, cybersecurity, finance, business, environmental issues, social considerations, and governance (ESG). This study underscores a systematic approach to understanding and managing the interaction between AI policy and the ICT-energy nexus regarding energy balance, supply chains, human factors, and regulation. Given today’s complex challenges, and being informed by best practices and future strategies that foster sustainable and resilient data centers and AI, the author argues that any solution should be threat-agnostic to ensure resilience against man-made and natural disasters. This interdisciplinary strategy ultimately aims to enhance energy efficiency, power transmission, grid availability, and supply chain resilience while reducing carbon emissions for ICT-energy systems. The state-of-the-art analysis identified gaps in policy papers and produced key findings to bolster AI capabilities. Focusing on Ukraine, the author presents these key findings as a pathway for further assumptions and research to highlight the elements of collaboration in policy, technology, human-centered approaches, and interdisciplinary efforts, coupled with a call to action.