Experts advocate open-source AI for safer military use

Experts at the Singapore Defence Technology Summit discussed the potential safety benefits of open-source AI models for military use. They argued that sharing source codes can help improve AI systems by involving a wider community. Rodrigo Liang, CEO of SambaNova Systems, noted that many think closed-source models are safer. However, he believes this is a misconception. He pointed out that bad actors can access closed-source models, which does not necessarily keep them safer. Pascale Fung, a professor at the Hong Kong University of Science and Technology, stressed the importance of open-source models, especially for national security. She mentioned that open-source models can be inspected and reviewed. By sharing the source codes, researchers can work together to enhance the safety and effectiveness of these AI systems.